Jui. Aug. 2017 Sep. Sun Mon Tue Wed Thu Fri Sat 30 31 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 01 02 □ Click Seminar

Past Seminars

DateAugust 21, 2017 (Monday)
SpeakerYutaro Yamaguchi
TitleStochastic Packing Integer Programs with Few Queries
SpeakerYuya Yoshikawa
Title
DateAugust 2, 2017 (Wednesday)
SpeakerXuemin Lin
TitleAdvances in Big Graph Processing
DateJuly 28, 2017 (Friday)
SpeakerJunpei Komiyama
TitleStatistical Emerging Pattern Mining with Multiple Testing Correction
DateJuly 27, 2017 (Thursday)
SpeakerMasaaki Imaizumi
Title
DateJuly 24, 2017 (Monday)
SpeakerShiho Sugimoto
TitleComputing Abelian string regularities based on RLE
DateJuly 18, 2017 (Tuesday)
SpeakerHara Satoshi
Title
DateJuly 10, 2017 (Monday)
SpeakerHayashi Kaede
Title
DateJuly 3, 2017 (Monday)
SpeakerTasuku Soma
TitleRegret Ratio Minimization in Multi-objective Submodular Function Maximization
DateJune 29, 2017 (Thursday)
SpeakerAkihiro Kishimoto
TitleEfficient AND/OR search algorithms for exact MAP inference task over graphical models
DateJune 22, 2017 (Thursday)
SpeakerRyuhei Uehara
Title
DateJune 5, 2017 (Monday)
SpeakerAndrea Marino
TitleEnumeration Algorithms: Introduction and Techniques
SpeakerRoberto Grossi
TitleNetwork Analytics via Pattern Discovery
DateMay 15, 2017 (Monday)
SpeakerYutaka Nagahata
Title
DateApril 10, 2017 (Monday)
SpeakerShinsaku Sakaue
Title
DateApril 7, 2017 (Friday)
SpeakerShin-ichi Minato
Title
DateFebruary 28, 2017 (Tuesday)
SpeakerKazuto FUKUCHI
TitleTBA
DateFebruary 24, 2017 (Friday)
SpeakerMarie KATSURAI
TitleTBA
DateFebruary 21, 2017 (Tuesday)
SpeakerTakanori MAEHARA
TitleTBA
SpeakerKazuki YOSHIZOE
TitleTBA
DateFebruary 16, 2017 (Thursday)
SpeakerYuu Jinnai
Title
DateFebruary 9, 2017 (Thursday)
SpeakerNaoto OSAKA
TitleTBA
DateJanuary 30, 2017 (Monday)
SpeakerJunya HONDA
TitleTBA
SpeakerJunpei KOMIYAMA
TitleTBA
DateJanuary 24, 2017 (Tuesday)
SpeakerNanao KITA
TitleTBA
DateNovember 25, 2016 (Friday)
SpeakerToru Mano
TitleEfficient Virtual Network Optimization Across Multiple Domains Without Revealing Private Information
DateOctober 31, 2016 (Monday)
SpeakerShinya TAKAMAEDA
Title
DateSeptember 29, 2016 (Thursday)
SpeakerTakanori MAEHARA
Title
SpeakerAtsuki Nagao
TitleTBA
DateSeptember 28, 2016 (Wednesday)
SpeakerShanfeng Zhu
TitleDeepMeSH and MeSHLabeler: Recent progress in large-scale MeSH indexing
DateSeptember 12, 2016 (Monday)
SpeakerJunpei Komiyama
TitleTBA
DateAugust 30, 2016 (Tuesday)
SpeakerKengo Nakamura
Title
DateAugust 22, 2016 (Monday)
SpeakerHei Chan
TitleConverting naive Bayes classifiers into OBDDs and other data structures for knowledge representation
DateJuly 8, 2016 (Friday)
SpeakerMasakazu Ishihata
Title
DateJuly 5, 2016 (Tuesday)
SpeakerKazuki Yoshizoe
Title
DateJune 13, 2016 (Monday)
SpeakerKunihiro Wasa
TitleTBA
DateMay 31, 2016 (Tuesday)
SpeakerHiromi Arai
TitleTBA
DateMay 23, 2016 (Monday)
SpeakerShogo Takeuchi
TitleTBA
DateMay 17, 2016 (Tuesday)
SpeakerMathias Soeken
TitleAncilla-free reversible logic synthesis using symbolic methods
DateFebruary 22, 2016 (Monday)
SpeakerYoshio Okamoto
Title
DateFebruary 16, 2016 (Tuesday)
SpeakerSatoshi Morinaga and Yuichi Nakamura
Title
DateJanuary 28, 2016 (Thursday)
SpeakerTakashi Washio
Title
DateJanuary 19, 2016 (Tuesday)
SpeakerHei Chan
TitleIdentifying Causal Effects in Linear SEMs
DateDecember 25, 2015 (Friday)
SpeakerYasuyuki Shirai
Title
DateDecember 7, 2015 (Monday)
SpeakerSadamori Koujaku
TitleTBA
DateDecember 3, 2015 (Thursday)
SpeakerKaori KURIBAYASHI-SHIGETOMI
TitleTBA
DateNovember 30, 2015 (Monday)
SpeakerYuto Nakashima
TitleTBA
DateOctober 23, 2015 (Friday)
SpeakerEndre Boros
TitleGenerating Bodies, Simplices, and Vertices of Polyhedra
DateOctober 23, 2015 (Friday)
SpeakerKoji Tuda
Title
DateOctober 20, 2015 (Tuesday)
SpeakerKatsuhiko HAYASHI
Title
DateSeptember 29, 2015 (Tuesday)
SpeakerKeigo Kimura
Title
DateSeptember 3, 2015 (Thursday)
SpeakerHiroyuki Namba
TitleOn Shortest Disjoint Paths by Computing Permanents
SpeakerNaoki Marumo
Title
SpeakerShinsaku Sakaue
Title
DateSeptember 2, 2015 (Wednesday)
SpeakerSyuhei Denzumi
TitleAdvanced Hybrid DenseZDD and Recent Topics
SpeakerYuni Iwamasa
Title
DateAugust 31, 2015 (Monday)
SpeakerISHII, Toshimasa
Title
DateAugust 19, 2015 (Wednesday)
SpeakerJun Kawahara
Title
DateJuly 10, 2015 (Friday)
SpeakerHiroshi Sawada
Title
DateJuly 7, 2015 (Tuesday)
SpeakerKimihito Ito
TitleCombining Population Genetics of Pathogens and Epidemiology of Infectious Diseases
DateMay 15, 2015 (Friday)
SpeakerDavid Avis
TitlePolynomial size matching polytopes
DateMay 12, 2015 (Tuesday)
SpeakerTakeru Inoue
Title
DateApril 27, 2015 (Monday)
SpeakerRyoma Sinya
Title
DateFebruary 6, 2015 (Friday)
SpeakerMatias Korman
TitleExtremal distances in Polygonal Domains
DateJanuary 30, 2015 (Friday)
SpeakerDavid Avis
TitleA portable parallel implementation of the lrs vertex enumeration code
DateJanuary 16, 2015 (Friday)
SpeakerHiroki Nakahara
TitleTBA
DateDecember 25, 2014 (Thursday)
SpeakerSeika Abe
TitleTBD
DateDecember 19, 2014 (Friday)
SpeakerKazuhisa Makino
TitleTBD
SpeakerTakeaki Uno
TitleTBD
DateDecember 5, 2014 (Friday)
SpeakerShogo Takeuchi, Ryutaro Kurai
TitleIntroduction to SDD (Sentential Decision Diagram) and its software package
DateNovember 14, 2014 (Friday)
SpeakerYasuyuki Shirai
TitleUnderstanding Changes in Customer Purchase Behavior: Study of Attenuation Model for Multiparametric Purchase Preferences
DateOctober 30, 2014 (Thursday)
SpeakerTakashi Washio
TitleMass based estimation of data distribution and dissimilarity
DateOctober 24, 2014 (Friday)
SpeakerKoji Tuda
TitleTBA
DateOctober 3, 2014 (Friday)
SpeakerIsao Tanaka
TitleTBD
DateSeptember 3, 2014 (Wednesday)
SpeakerTaiji Suzuki
TitleTBA
DateAugust 25, 2014 (Monday)
SpeakerMichael Lampis
TitleParameterized Approximation Schemes using Graph Widths
DateAugust 19, 2014 (Tuesday)
SpeakerKazuyoshi Sasaki
TitleTBA
DateAugust 5, 2014 (Tuesday)
SpeakerShohei Hido
TitleTBA
DateJuly 15, 2014 (Tuesday)
SpeakerKitsuchart Pasupa
TitleCan relevance of images be inferred from eye movements?
DateJuly 7, 2014 (Monday)
SpeakerLeonardo Bartoloni
TitleTime/space tradeoff on Discrete Optimization using approximated Binary Decision Diagrams
DateJuly 4, 2014 (Friday)
SpeakerHiroshi Tohjo
TitleTBA
DateMay 28, 2014 (Wednesday)
SpeakerAtsushi Takizawa
TitleTBA
DateMay 7, 2014 (Wednesday)
SpeakerNaohide Nagatsu
TitleTBA
DateApril 30, 2014 (Wednesday)
SpeakerYutaka Nagahata
TitleTBA
DateApril 16, 2014 (Wednesday)
SpeakerValia Mitsou
TitleComputational complexity of games and puzzles, and, in particular of the card game SET
DateApril 11, 2014 (Friday)
SpeakerHiroyuki Hanada
TitleAverage-case linear-time similar substring searching by the q-gram distance
DateFebruary 26, 2014 (Wednesday)
SpeakerTakeru Inoue
TitleGraphillion update
SpeakerNorihito Yasuda
TitleCase studies in enumerating subtrees using Graphillion
SpeakerMasakazu Nakamoto
TitleEkillion as an Applications of Graphillion
SpeakerYuichiro Yasui
TitleTBA
DateJanuary 24, 2014 (Friday)
SpeakerHiroaki Iwashita
TitleTITLE TBA
DateJanuary 17, 2014 (Friday)
SpeakerTakahisa Toda
TitleDualization of Boolean functions Using Ternary Decision Diagrams
DateJanuary 10, 2014 (Friday)
SpeakerCharles Jordan
TitleTowards Large-Scale Experimental Descriptive Complexity
DateDecember 16, 2013 (Monday)
SpeakerPhillippe Samer
TitleA branch and cut algorithm for spanning trees under conflict constraints
DateDecember 10, 2013 (Tuesday)
SpeakerNorihito Yasuda
TitleTBA
DateDecember 6, 2013 (Friday)
SpeakerNabila Abdessaied
TitleUpper bounds for reversible circuits based on Young subgroups
DateDecember 3, 2013 (Tuesday)
SpeakerYoshihiko Suhara
TitleTBA
DateNovember 28, 2013 (Thursday)
SpeakerWill Archer Arentz
TitleWhat is Rakuten Institute of Technology (RIT)?
DateNovember 8, 2013 (Friday)
SpeakerYasuyuki Shirai
TitleTBA
DateNovember 1, 2013 (Friday)
SpeakerTakashi Washio
TitleTBA
DateOctober 28, 2013 (Monday)
SpeakerYasuo Tabei
TitleTBA
DateOctober 25, 2013 (Friday)
SpeakerKoji Tuda
TitleData Mining in Biological Science
DateOctober 8, 2013 (Tuesday)
SpeakerFlorian Lonsing
TitleSearch-based QBF Solving: Basics, Recent Trends and Challenges
DateOctober 1, 2013 (Tuesday)
SpeakerYutaka Nagahata
TitleCombination explosion problem in extraction of hierarchical structure from steady state Markov chain
DateSeptember 20, 2013 (Friday)
SpeakerRajeev Raman
TitleEncodings for Range Selection and Top-k Queries
DateSeptember 18, 2013 (Wednesday)
SpeakerShin-ichiro Suzuki
TitleMuseum and the Web 2013 Report
DateSeptember 12, 2013 (Thursday)
SpeakerYuko Sakurai
TitleTBA
DateSeptember 3, 2013 (Tuesday)
SpeakerNovi Quadrianto
TitleSimple and Efficient Bayesian Random Forest
DateAugust 29, 2013 (Thursday)
SpeakerMartina Seidl
TitleRecent Advancements in QBF Solving
DateAugust 28, 2013 (Wednesday)
SpeakerZhiyong Peng
TitleKey Management for Access Control in Trusted Cloud Storage
DateAugust 27, 2013 (Tuesday)
SpeakerTakuya Akiba
TitleTBA
DateAugust 26, 2013 (Monday)
SpeakerŁukasz Kaiser
TitleExperimental Descriptive Complexity and Machine Learning
DateAugust 8, 2013 (Thursday)
SpeakerHiroki Nakahara
TitleTBA
DateAugust 1, 2013 (Thursday)
SpeakerTakehiro Ito
TitleTBA
DateJuly 16, 2013 (Tuesday)
SpeakerShinobu Nagayama
TitleTBA
DateJuly 4, 2013 (Thursday)
SpeakerTakanori Maehara
TitleTBA
DateJuly 2, 2013 (Tuesday)
SpeakerMarco Cuturi
TitleLightspeed Optimal Transportation Distances for Histograms
DateMay 29, 2013 (Wednesday)
SpeakerNorihito Yasuda
TitleTBA
DateMay 9, 2013 (Thursday)
SpeakerFlorian Horn
TitleStochastic Games in Verification
DateApril 24, 2013 (Wednesday)
SpeakerMichael E. Houle
TitleIntrinsic Dimensionality and Discriminability of Data
DateApril 12, 2013 (Friday)
SpeakerKazuki Yoshizoe
TitleTBA
DateMarch 26, 2013 (Tuesday)
SpeakerMasanori Sugimoto
TitleTBA
SpeakerTetsuya Yoshida
TitleTBA
DateMarch 22, 2013 (Friday)
SpeakerMasakazu Iahihata
TitleProposionalized Probability Computation and Learning on Binary Decision Diagrams
DateMarch 21, 2013 (Thursday)
SpeakerMartin Müller
TitleMove Quality in Monte Carlo Simulation: A Case Study using the Fuego Go Program
DateMarch 11, 2013 (Monday)
SpeakerToshiki Saitoh
TitleTBA
DateMarch 9, 2013 (Saturday)
SpeakerHiroaki Iwashita, Yoshio Nakawaza
TitleTBA
DateJanuary 31, 2013 (Thursday)
SpeakerShin Miyakawa
TitleTBA
DateJanuary 28, 2013 (Monday)
SpeakerRobert Wille and Mathias Soeken
TitleSynthesis of Reversible Circuits using Decision Diagrams: Background,Recent Accomplishments, and Open Challenges
DateJanuary 15, 2013 (Tuesday)
SpeakerRyo Yoshinaka
TitleTBA
DateJanuary 11, 2013 (Friday)
SpeakerShin-ichi Minato
TitleTBA
DateDecember 12, 2012 (Wednesday)
SpeakerToshio endo
TitleTBA
SpeakerHitoshi Sato
TitleTBA
SpeakerYuichiro Yasui
TitleTBA
SpeakerKatsuki Fujisawa
TitleTBA
DateDecember 6, 2012 (Thursday)
SpeakerShoji Kasahara
TitleTBA
DateDecember 5, 2012 (Wednesday)
SpeakerRajeev Raman
TitleDynamizing Succinct Tree Representations
DateDecember 5, 2012 (Wednesday)
SpeakerRoberto Grossi
TitleEmpowering Succinct Data Structures for Strings and Sequences
DateNovember 22, 2012 (Thursday)
SpeakerAlexandre Termier
TitleParaMiner: generic and parallel pattern mining scaling to real-world data
DateNovember 19, 2012 (Monday)
SpeakerNorie Fu
TitleTBA
DateNovember 14, 2012 (Wednesday)
SpeakerKohei Hayashi
TitleTBA
DateNovember 12, 2012 (Monday)
SpeakerS.R.K. Branavan
TitleGrounding Linguistic Analysis in Control Applications
DateNovember 9, 2012 (Friday)
SpeakerYasuyuki Shirai
TitleSet-Similarity Joins
DateNovember 6, 2012 (Tuesday)
SpeakerTakahisa Toda
TitleOn Separating Convex Points with Lines
DateNovember 2, 2012 (Friday)
SpeakerTakashi Washio
TitleGlobal Maximization of Submodular Function and Its application to Machine Learning
DateOctober 26, 2012 (Friday)
SpeakerKoji Tsuda
TitleTBA
DateOctober 3, 2012 (Wednesday)
SpeakerLukasz Kaiser
TitleLearning and Playing Board Games using Descriptive Complexity
DateSeptember 14, 2012 (Friday)
SpeakerHidetomo Nabeshima
TitleThe principles of the high performance SAT solvers
DateSeptember 13, 2012 (Thursday)
SpeakerSeiichiro Tani
TitleTBA
DateSeptember 12, 2012 (Wednesday)
SpeakerProf. Rusins Freivalds
TitleUltrametric automata and Turing machines
DateSeptember 7, 2012 (Friday)
SpeakerTakashi Horiyama
TitleTBA
DateAugust 10, 2012 (Friday)
SpeakerYoshihiro Yamanishi
TitleMachine learning methods to infer drug-target interaction networks
DateAugust 3, 2012 (Friday)
SpeakerShin Aida
TitleTBA
DateJuly 6, 2012 (Friday)
SpeakerMutsunori Banbara
TitleTBA
DateJune 8, 2012 (Friday)
SpeakerIchigaku Takigawa
TitleTBA
DateMay 30, 2012 (Wednesday)
SpeakerTakeshi Yamamuro
TitleCompression and Searches with Modern Processors (Practice and Implementation )
DateMay 30, 2012 (Wednesday)
SpeakerMakoto Onizuka
Title
DateMay 24, 2012 (Thursday)
SpeakerShogo Takeuchi
Title
DateApril 20, 2012 (Friday)
SpeakerHiroshi Aoki
TitleTBA
DateApril 13, 2012 (Friday)
SpeakerKensuke Fukuda
Title
DateMarch 30, 2012 (Friday)
SpeakerShi-ichiro Suzuki
TitleTBA
DateFebruary 24, 2012 (Friday)
SpeakerKazuki Yoshizoe
TitleTBA
DateFebruary 20, 2012 (Monday)
SpeakerMasakazu Iahihata
TitleBayesian networks and BDDs
DateFebruary 10, 2012 (Friday)
SpeakerCharles Jordan
TitleDescriptive Programming
DateJanuary 20, 2012 (Friday)
SpeakerTetsuo Shibuya
Title
DateDecember 26, 2011 (Monday)
SpeakerTakahisa Toda
TitleTBA
DateDecember 19, 2011 (Monday)
SpeakerKeisuke Hotta
TitleTBA
DateNovember 25, 2011 (Friday)
SpeakerTakashi Washio
Title
DateNovember 24, 2011 (Thursday)
SpeakerYasuyuki Shirai
TitleIntroduction to data mining application using large-scale data
DateNovember 14, 2011 (Monday)
SpeakerShingo Yoshizawa
Title
DateOctober 27, 2011 (Thursday)
SpeakerTetsuo Asano
TitleDesigning Algorithms with Limited Work Space
DateOctober 14, 2011 (Friday)
SpeakerKoji Tuda
TitleUltra-fast algorithms for all pairs similarity search and their application to biological data.
DateOctober 12, 2011 (Wednesday)
SpeakerEtsuji Tomita
Title
DateSeptember 30, 2011 (Friday)
SpeakerEishi Chiba
Title
DateSeptember 14, 2011 (Wednesday)
SpeakerHirokazu Anai
Title
DateAugust 30, 2011 (Tuesday)
SpeakerJun Sakuma
Title
DateAugust 29, 2011 (Monday)
SpeakerTakashi Horiyama
Title
DateAugust 26, 2011 (Friday)
SpeakerNaoki Kato
Title
DateAugust 25, 2011 (Thursday)
SpeakerHiroki Nakahara
TitleTBD
DateAugust 25, 2011 (Thursday)
SpeakerAkihiro Kishimoto
TitleDepth-First Proof-Number Search in the Presence of Repetitions
DateAugust 24, 2011 (Wednesday)
SpeakerKunihiko Sadakane
TitleTBA
DateAugust 23, 2011 (Tuesday)
SpeakerMasashi Sugiyama
Title
DateAugust 23, 2011 (Tuesday)
SpeakerHiroo Ito
TitleTBA
DateAugust 9, 2011 (Tuesday)
SpeakerTakeaki Uno
Title
DateJuly 29, 2011 (Friday)
SpeakerSeigo Takeuchi
TitleTBA
DateJuly 15, 2011 (Friday)
SpeakerMasaaki Nishino
Title
DateJuly 8, 2011 (Friday)
SpeakerKazutaka Matsuzaki
Title
DateJuly 1, 2011 (Friday)
SpeakerShin-ichi Minato
Title
DateJune 17, 2011 (Friday)
SpeakerHiroaki Iwashita
TitleGraph Path Enumeration by ZDD-Mate Method
DateMay 24, 2011 (Tuesday)
SpeakerProf. Neil Immerman
Title"P versus NP: Approaches, Rebuttals, and Does It Matter?"
DateMay 20, 2011 (Friday)
SpeakerJun Kawahara
Title
DateApril 21, 2011 (Thursday)
SpeakerShin-ichi Minato
Title
DateApril 15, 2011 (Friday)
SpeakerShunji Umetani
Title
DateApril 8, 2011 (Friday)
SpeakerMichael Houle
TitleIntrinsic Dimensionality and its Applications to Databases and Data Mining
DateMarch 23, 2011 (Wednesday)
SpeakerShin-ichi Minato
Title
DateMarch 22, 2011 (Tuesday)
SpeakerShinichi Minato
Title
DateMarch 7, 2011 (Monday)
Speaker Dr. Benjamin Rossman
TitleAverage-case complexity of detecting cliques
DateFebruary 28, 2011 (Monday)
SpeakerRyutaro Kurai
Title
DateFebruary 24, 2011 (Thursday)
SpeakerTsuyoshi Ueno
Title
DateFebruary 7, 2011 (Monday)
SpeakerDr. Paula Brito
TitleIntroduction to Symbolic Data Analysis (An interaction movement between statistics and data processing)
DateJanuary 31, 2011 (Monday)
SpeakerDr. Mathias Soeken
TitleFormal Verification of UML-based Specifications
DateJanuary 28, 2011 (Friday)
SpeakerHiroaki Iwashita
TitleImproving Simulation Coverage of Metastability Effects in Clock Domain　Crossing Verification
DateJanuary 21, 2011 (Friday)
SpeakerMasaaki　Kotera
Title
DateJanuary 11, 2011 (Tuesday)
SpeakerMakoto Kanazawa
TitleA Chomsky-Sch\"{u}tzenberger-Weir Representation Theorem for Simple Context-Free Tree Grammars
DateJanuary 7, 2011 (Friday)
SpeakerKimihito Ito
TitlePrediction of amino acid substitutions on the hemagglutinin　molecules of influenza A viruses
DateDecember 17, 2010 (Friday)
SpeakerGuan-Cheng Li
TitleMining Psychology from English News and Applications on Finance.
DateNovember 19, 2010 (Friday)
SpeakerProf.　Kai Ming Ting
TitleMulti-Dimensional Mass Estimation and Mass-based Clustering
DateNovember 15, 2010 (Monday)
SpeakerProf. Adnan Darwiche
TitleEfficient Representations of Boolean Functions: The View from Knowledge Compilation
DateNovember 12, 2010 (Friday)
SpeakerJoe Suzuki
Title
DateNovember 5, 2010 (Friday)
SpeakerRyo Yoshinaka
TitleIntroduction to Sequence BDD and Comparison with Previous Methods
SpeakerShuhei Denzumi
TitleSeqBDD : Introduction to Sequence BDD
DateOctober 29, 2010 (Friday)
SpeakerTakeru Inoue
Title
DateOctober 22, 2010 (Friday)
SpeakerShin-ichi　Minato
Title
DateSeptember 27, 2010 (Monday)
SpeakerShohei Hido
Title
DateSeptember 27, 2010 (Monday)
SpeakerTakashi Imamichi
Title
DateSeptember 27, 2010 (Monday)
SpeakerTsuyoshi　Ide
Title
DateSeptember 16, 2010 (Thursday)
SpeakerTakashi Horiyama
Title
DateSeptember 2, 2010 (Thursday)
SpeakerProf. Randy Goebel
TitleChallenges for a theory of visualization: what is semantic symmetry?
DateAugust 26, 2010 (Thursday)
SpeakerYasuo Tabei
Title
DateAugust 23, 2010 (Monday)
SpeakerTsuyoshi　Shibata
Title
DateAugust 20, 2010 (Friday)
SpeakerShigeru Yamashita
Title
DateAugust 20, 2010 (Friday)
SpeakerByung-Soo Choi
Title
DateAugust 6, 2010 (Friday)
SpeakerTakeaki Uno
Title
DateAugust 6, 2010 (Friday)
SpeakerAkihiro Yamamoto
Title
DateAugust 4, 2010 (Wednesday)
SpeakerRyuhei Uehara
Title
DateAugust 3, 2010 (Tuesday)
SpeakerShigeru Yamashita
TitleIntroduction to Grover Algorithm
DateAugust 3, 2010 (Tuesday)
SpeakerProf. Byung-Soo Choi
TitleGrover Search and its Applications
DateJuly 23, 2010 (Friday)
SpeakerTakuya Kida
Title
DateJuly 9, 2010 (Friday)
SpeakerYasuyuki Shirai
Title
DateJuly 2, 2010 (Friday)
SpeakerToshiki Saitoh
Title
DateJune 18, 2010 (Friday)
SpeakerJun Kawahara
Title
DateJune 4, 2010 (Friday)
SpeakerRyo Yoshinaka
TitleChomsky-Schützenberger-Type Characterization of Multiple Context-Free Languages
DateMay 21, 2010 (Friday)
SpeakerShin-ichi　Minato
Title
DateMay 14, 2010 (Friday)
SpeakerZin Arai
Title

Detail

DateAugust 21, 2017 (Monday)
Site
Time15:00－16:00
SpeakerYutaro Yamaguchi
ProfileAssistant Prof. at Osaka University
TitleStochastic Packing Integer Programs with Few Queries
AbstractWe consider a stochastic variant of the integer linear programming problem, which contains random variables in the objective vector. We are allowed to reveal each entry of the objective vector by conducting a query, and the task is to find a good solution by conducting a small number of queries. We propose adaptive and non-adaptive algorithms for this problem, and provide a general technique for analyzing the performance of the algorithms. We also demonstrate our framework by applying it to a variety of stochastic combinatorial optimization problems such as matching, matroid, and stable set problems. This is a joint work with Takanori Maehara in RIKEN AIP.
Time16:00－17:00
SpeakerYuya Yoshikawa
Profile
Title
Abstract

Detail

DateAugust 2, 2017 (Wednesday)
Site
Time10:30－11:30
SpeakerXuemin Lin
ProfileScientia Professor, The University of New South Wales
TitleAdvances in Big Graph Processing
AbstractGraphs are very important parts of Big Data and widely used for modelling complex structured data with a broad spectrum of applications such as bioinformatics, web search, social network, road network, etc. Over the last decade, tremendous research efforts have been devoted to many fundamental problems in managing and analysing graph data. In this talk, I will first overview our recent research efforts in processing big graphs including scalable processing theory and techniques, distributed computation, and system framework. We will also look to the future of the area.

Detail

DateJuly 28, 2017 (Friday)
Site
Time10:30－11:30
SpeakerJunpei Komiyama
Profile
TitleStatistical Emerging Pattern Mining with Multiple Testing Correction
AbstractEmerging patterns are patterns whose support significantly differs between two databases. We study the problem of listing emerging patterns with a multiple testing guarantee. Recently, Terada et al. proposed the Limitless Arity Multiple-testing Procedure (LAMP) that controls the family-wise error rate (FWER) in statistical association mining. LAMP reduces the number of untestable’’ hypotheses without compromising its statistical power. Still, FWER is restrictive, and as a result, its statistical power is inherently unsatisfying when the number of patterns is large.

On the other hand, the false discovery rate (FDR) is less restrictive than FWER, and thus controlling FDR yields a larger number of significant patterns. We propose two emerging pattern mining methods: the first one controls FWER, and the second one controls FDR. The effectiveness of the methods is verified in computer simulations with real-world datasets.

Detail

DateJuly 27, 2017 (Thursday)
Site
Time13:00－14:00
SpeakerMasaaki Imaizumi
Profile
Title
AbstractWe analyze the statistical efficiency of the probability density estimation problem when the density function is highly non-smooth. The problem of density estimation appears in various situations, and significantly affects statistics and machine learning. In the existing studies, smoothness of density functions is necessary to measure the statistical efficiency of the estimation. By contrast, the estimation of non-smooth density functions remains an open question, although the non-smooth density functions frequently appear in the application fields. In this paper, we propose a Szemeredi density estimator (SDE) which is an estimator for non-smooth density functions based on graph theory. We derive the speed of convergence of SDE, then clarify the statistical efficiency for the estimation of non-smooth densities. Furthermore, we discuss optimal and adaptive properties of SDE, and experiments to verify the efficiency results of SDE.
This is joint work with Takanori Maehara (RIKEN).

Detail

DateJuly 24, 2017 (Monday)
Site
Time16:00－17:00
SpeakerShiho Sugimoto
Profile
TitleComputing Abelian string regularities based on RLE
AbstractTwo strings x and y are said to be Abelian equivalent if x is a permutation of y, or vice versa. If a string z satisfies z = xy with x and y being Abelian equivalent, then z is said to be an Abelian square. If a string w can be factorized into a sequence v1,...,vs of strings such that v1 , . . . , vs−1 are all Abelian equivalent and vs is a substring of a permutation of v1, then w is said to have a regular Abelian period (p,t) where p = |v1| and t = |vs|. If a substring w1[i..i+l−1] of a string w1 and a substring w2[j..j + l − 1] of another string w2 are Abelian equivalent, then the substrings are said to be a common Abelian factor of w1 and w2 and if the length l is the maximum of such then the substrings are said to be a longest common Abelian factor of w1 and w2. We propose efficient algorithms which compute these Abelian regularities using the run length encoding (RLE) of strings. For a given string w of length n whose RLE is of size m, we propose algorithms which compute all Abelian squares occurring in w in O(mn) time, and all regular Abelian periods of w in O(mn) time. For two given strings w1 and w2 of total length n and of total RLE size m, we propose an algorithm which computes all longest common Abelian factors in O(m2n) time.

Detail

DateJuly 18, 2017 (Tuesday)
Site
Time15:00－16:00
SpeakerHara Satoshi
Profile
Title
Abstract

Detail

DateJuly 10, 2017 (Monday)
Site
Time16:00－17:00
SpeakerHayashi Kaede
Profile
Title
Abstract

Detail

DateJuly 3, 2017 (Monday)
Site
Time13:30－14:30
SpeakerTasuku Soma
ProfileThe University of Tokyo
TitleRegret Ratio Minimization in Multi-objective Submodular Function Maximization
AbstractSubmodular function maximization has numerous applications in machine learning and artificial intelligence. Many real applications require multiple submodular objective functions to be maximized, and it is not known in advance which of the objective functions is regarded to be important by a user. In such cases, it is desirable to have a small family of representative solutions that would satisfy any user’s preference. A traditional approach for solving such a problem is to enumerate the Pareto optimal solutions. However, owing to the massive number of Pareto optimal solutions (possibly exponentially many), it is difficult for a user to select a solution. In this paper, we propose practical methods with theoretical guarantees for finding a small family of representative solutions, based on the notion of regret ratio. This is joint work with Yuichi Yoshida.

Detail

DateJune 29, 2017 (Thursday)
Site
Time10:00－11:00
SpeakerAkihiro Kishimoto
ProfileResearch Staff Member, IBM Ireland
TitleEfficient AND/OR search algorithms for exact MAP inference task over graphical models
AbstractGraphical models provide a powerful framework for reasoning with probabilistic information. Combinatorial maximization, or maximum a posteriori (MAP) tasks arise in many applications and often can be efficiently solved by search schemes, especially in the context of AND/OR search spaces that are sensitive to the underlying problem structure.
In this talk, I present the power of limited memory best-first search over AND/OR search spaces, named RBFAOO, which performs exact MAP inference over graphical models. I also present parallelized versions of RBFAOO which run in both shared-memory and distributed-memory environments. I show that RBFAOO is empirically superior to the current state-of-the-art approaches based on AND/OR search, especially on very hard problem instances.

Detail

DateJune 22, 2017 (Thursday)
Site
Time13:30－14:30
SpeakerRyuhei Uehara
Profile
Title
Abstract

Detail

DateJune 5, 2017 (Monday)
Site
Time15:00－16:00
SpeakerAndrea Marino
ProfileAssistant Professor at University of Pisa
TitleEnumeration Algorithms: Introduction and Techniques
AbstractThe development of algorithms for enumerating all possible solutions
of a specific combinatorial problem has a long history: already in the
1960s the problem of enumerating some specific graph-theoretic structures
(such shortest paths and cycles) has been attacked. This new interest
towards graph enumeration comes from the analysis of biological (and, more
in general, of complex) networks. The main goal of this talk is revising
the basics of graph enumeration, summarising the efficiency measures and
the main current available algorithm design techniques.
Time16:00－17:00
SpeakerRoberto Grossi
ProfileProfessor of Algorithms and Data Structures at the University of Pisa
TitleNetwork Analytics via Pattern Discovery
AbstractSocial, biological and communication networks of data with a strong linked
nature can be modeled and analyzed as labeled graphs. We describe some new
algorithms for pattern discovery in graphs that can be useful for network
analytics, focussing on clique enumeration and its applications.

Joint work with A. Conte, A. Marino, L. Tattini, and L. Versari

Detail

DateMay 15, 2017 (Monday)
Site
Time14:00－15:00
SpeakerYutaka Nagahata
Profile
Title
Abstract

Detail

DateApril 10, 2017 (Monday)
Site
Time16:00－17:30
SpeakerShinsaku Sakaue
Profile
Title
Abstract

Detail

DateApril 7, 2017 (Friday)
Site
Time16:00－17:30
SpeakerShin-ichi Minato
Profile
Title
Abstract

Detail

DateFebruary 28, 2017 (Tuesday)
Site
Time16:00－17:00
SpeakerKazuto FUKUCHI
Profile
TitleTBA
AbstractTBA

Detail

DateFebruary 24, 2017 (Friday)
Site
Time16:00－17:00
SpeakerMarie KATSURAI
Profile
TitleTBA
AbstractTBA

Detail

DateFebruary 21, 2017 (Tuesday)
Site
Time15:00－16:00
SpeakerTakanori MAEHARA
Profile
TitleTBA
AbstractTBA
Time16:00－17:00
SpeakerKazuki YOSHIZOE
Profile
TitleTBA
AbstractTBA

Detail

DateFebruary 16, 2017 (Thursday)
Site
Time16:00－17:00
SpeakerYuu Jinnai
Profile
Title
Abstract

Detail

DateFebruary 9, 2017 (Thursday)
Site
Time11:00－12:00
SpeakerNaoto OSAKA
Profile
TitleTBA
Abstract

Detail

DateJanuary 30, 2017 (Monday)
Site
Time15:00－16:00
SpeakerJunya HONDA
Profile
TitleTBA
AbstractTBA
Time16:00－17:00
SpeakerJunpei KOMIYAMA
Profile
TitleTBA
AbstractTBA

Detail

DateJanuary 24, 2017 (Tuesday)
Site
Time10:00－11:30
SpeakerNanao KITA
Profile
TitleTBA
AbstractTBA

Detail

DateNovember 25, 2016 (Friday)
Site
Time14:45－16:15
SpeakerToru Mano
Profile
TitleEfficient Virtual Network Optimization Across Multiple Domains Without Revealing Private Information
Abstract

Detail

DateOctober 31, 2016 (Monday)
Site
Time16:30－17:30
SpeakerShinya TAKAMAEDA
Profile
Title
Abstract

Detail

DateSeptember 29, 2016 (Thursday)
Site
Time10:30－11:30
SpeakerTakanori MAEHARA
Profile
Title
Abstract
Time16:30－17:30
SpeakerAtsuki Nagao
Profile
TitleTBA
Abstract

Detail

DateSeptember 28, 2016 (Wednesday)
Site
Time16:30－17:30
SpeakerShanfeng Zhu
Profile
TitleDeepMeSH and MeSHLabeler: Recent progress in large-scale MeSH indexing
AbstractMotivation: Medical Subject Headings (MeSHs) are used by National Library of Medicine (NLM) to index almost all citations in MEDLINE, which greatly facilitates the applications of biomedical information retrieval and text mining. To reduce the time and financial cost of manual annotation, NLM has developed a software package, Medical Text Indexer (MTI), for assisting MeSH annotation, which uses k-nearest neighbors (KNN), pattern matching and indexing rules. Large-scale MeSH indexing has two challenging aspects: the MeSH side and the citation side. For the MeSH side, there are more than 27,000 distinct MeSH headings, and their distribution is biased. In addition, there are large variations in the number of MeSH for each citation. For the citation side, the existing methods mainly deal with text by bag-of-words, which cannot capture semantic and context-dependent information well.

Method: To address the challenge in the MeSH side, we propose a novel framework, MeSHLabeler, to integrate multiple evidence for accurate MeSH annotation by using 'learning to rank'. Evidence includes numerous predictions from MeSH classifiers, KNN, pattern matching, MTI and the correlation between different MeSH terms, etc. We further propose DeepMeSH that incorporates deep semantic information for large-scale MeSH indexing. The citation side challenge is solved by a new deep semantic representation, D2V-TFIDF, which concatenates both sparse and dense semantic representations.

Result: DeepMeSH achieved a Micro F-measure of 0.6323, 2% higher than 0.6218 of MeSHLabeler and 12% higher than 0.5637 of MTI, for BioASQ3 challenge data with 6000 citations. In addition, DeepMeSH achieved the first place in the BioASQ4 challenge, and MeSHLabeler achieved first place in both BioASQ2 and BioASQ3 challenges.

Reference:
Shengwen Peng, Ronghui You, Hongning Wang, Chengxiang Zhai, Hiroshi Mamitsuka, Shanfeng Zhu:
DeepMeSH: deep semantic representation for improving large-scale MeSH indexing. Bioinformatics (ISMB2016) 32(12): 70-79 (2016)
Ke Liu, Shengwen Peng, Junqiu Wu, ChengXiang Zhai, Hiroshi Mamitsuka, Shanfeng Zhu:
MeSHLabeler: improving the accuracy of large-scale MeSH indexing by integrating diverse evidence. Bioinformatics (ISMB2015) 31(12): 339-347 (2015)

Detail

DateSeptember 12, 2016 (Monday)
Site
Time16:30－17:30
SpeakerJunpei Komiyama
ProfileResearch Associate at the University of Tokyo
TitleTBA
AbstractTBA

Detail

DateAugust 30, 2016 (Tuesday)
Site
Time16:30－17:30
SpeakerKengo Nakamura
Profile
Title
Abstract

Detail

DateAugust 22, 2016 (Monday)
Site
Time16:30－17:30
SpeakerHei Chan
Profile
TitleConverting naive Bayes classifiers into OBDDs and other data structures for knowledge representation
AbstractBayesian network classifiers are used in many fields, and one common type of classifiers are naive Bayes classifiers, which contains a class variable as the root and a set of evidence variables as its children. In this talk, we introduce an approach for reasoning about Bayesian network classifiers in which we explicitly convert them into Ordered Binary Decision Diagrams (OBDDs). Specifically, we present an algorithm for converting any naive Bayes classifier into an OBDD, and we show theoretically and experimentally that this algorithm can give us an OBDD that is tractable in size even given an intractable number of instances. Since OBDDs are tractable representations of classifiers, our algorithm allows us to reason about the properties of these classifiers, such as to efficiently test the equivalence of two naive Bayes classifiers and characterize discrepancies between them. We also discuss extending our algorithm to converting naive Bayes classifiers into other data structures for knowledge representation, such as Zero-suppressed Decision Diagrams (ZDDs) and Sentential Decision Diagrams (SDDs).

Detail

DateJuly 8, 2016 (Friday)
Site
Time16:00－17:00
SpeakerMasakazu Ishihata
Profile
Title
Abstract

Detail

DateJuly 5, 2016 (Tuesday)
Site
Time16:30－17:30
SpeakerKazuki Yoshizoe
Profile
Title
Abstract

Detail

DateJune 13, 2016 (Monday)
SiteVBL 301B
Time16:30－17:30
SpeakerKunihiro Wasa
ProfileNational Institute of Informatics
TitleTBA
AbstractTBA

Detail

DateMay 31, 2016 (Tuesday)
SiteVBL 301B
Time13:00－14:00
SpeakerHiromi Arai
ProfileUniversity of Tokyo
TitleTBA
AbstractTBA

Detail

DateMay 23, 2016 (Monday)
SiteVBL 301B
Time14:30－15:30
SpeakerShogo Takeuchi
ProfileTokyo University
TitleTBA
AbstractTBA
Connected siteKanda Lab.

Detail

DateMay 17, 2016 (Tuesday)
SiteVBL 301B
Time16:30－17:30
SpeakerMathias Soeken
ProfileSwiss Federal Institute of Technology in Lausanne (EPFL)
TitleAncilla-free reversible logic synthesis using symbolic methods

AbstractDue to the properties of reversibility, reversible circuit synthesis that is optimum in the number of lines is a difficult task. For an irreversible Boolean function it is coNP-hard to find an optimum embedding, i.e., a reversible function with the minimum number of additional lines. Synthesis algorithms exist that obtain from an optimum embedding ancilla-free reversible circuits which have as many circuit lines as variables in the reversible function. However, so far all implementations for such synthesis algorithms require exponential time and space since they operate on the truth table representation of the function. In the talk, alternative implementations of the algorithms based on decision diagrams and Boolean satisfiability are presented that allow to run the algorithm in less space and time for some functions. It is shown that high quality synthesis results for large circuits can be obtained using these implementations.

Detail

DateFebruary 22, 2016 (Monday)
Site
Time15:30－17:00
SpeakerYoshio Okamoto
Profile
Title
Abstract

Detail

DateFebruary 16, 2016 (Tuesday)
Site
Time15:30－17:00
SpeakerSatoshi Morinaga and Yuichi Nakamura
Profile
Title
Abstract

Detail

DateJanuary 28, 2016 (Thursday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerTakashi Washio
Profile
Title
Abstract

Detail

DateJanuary 19, 2016 (Tuesday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerHei Chan
Profile
TitleIdentifying Causal Effects in Linear SEMs

AbstractIn this talk, we discuss the problem of identifying causal effects in linear Structural Equation Models (SEMs). Given the causal diagram of the SEM, which is a directed acyclic graph (with bi-directed edges to represent confounders), we want to find the solution of a causal effect from the observed covariances. The two methods for solving this problem are algebraic methods, which attempts to solve the set of simultaneous equations based on Wright's method of path analysis, and graphical methods, which tests for certain conditions in the causal diagram based on Pearl's d-separation test. To unify the two sets of methods, we introduce the Instrumental Variable Function, which we show to correspond to both algebraic and graphical methods, and is useful in helping us to solve this problem in a variety of graphs.

Detail

DateDecember 25, 2015 (Friday)
Site
Time15:30－17:00
SpeakerYasuyuki Shirai
Profile
Title
Abstract

Detail

DateDecember 7, 2015 (Monday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerSadamori Koujaku
Profile
TitleTBA
Abstract

Detail

DateDecember 3, 2015 (Thursday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time10:30－12:00
SpeakerKaori KURIBAYASHI-SHIGETOMI
Profile
TitleTBA
Abstract

Detail

DateNovember 30, 2015 (Monday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerYuto Nakashima
Profile
TitleTBA
Abstract

Detail

DateOctober 23, 2015 (Friday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time10:30－12:00
SpeakerEndre Boros
ProfileRutgers Univ.
TitleGenerating Bodies, Simplices, and Vertices of Polyhedra
AbstractMonotone generation problems are pervasive and are the underlying reason for the efficiency or inefficiency of many of the “large data” related tasks. Many of these problems have natural geometric representations, and lead to interesting and sometimes surprising connections. In this talk, we survey the complexities of generation problems and related geometric questions: given a finite set of points in an Euclidean space, what are the minimal subsets that contain a given point in their convex hull (simplices), in the interior of their convex hull (bodies), or the maximal counterparts of these. Analogously, given a set of linear inequalities, what are the minimal infeasible subsystems, or what are the maximal feasible subsystems? Both of these types of problems have close relations to the problem of generating the vertices of polyhedra represented as finite systems of linear inequalities, as well as to some problems on graphs and directed graphs. In this talk we survey these results and problems, their connections, and recall some open problems. (Based on joint research with K. Borys, K. Elbassioni, V. Gurvich and L. Khachiyan.)

Detail

DateOctober 23, 2015 (Friday)
Site
Time15:30－17:00
SpeakerKoji Tuda
Profile
Title
Abstract

Detail

DateOctober 20, 2015 (Tuesday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerKatsuhiko HAYASHI
ProfileNTT Communication Science Labs., NTT Japan.
Title
Abstract

Detail

DateSeptember 29, 2015 (Tuesday)
Site
Time15:30－17:00
SpeakerKeigo Kimura
Profile
Title
Abstract

Detail

DateSeptember 3, 2015 (Thursday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:00－16:00
SpeakerHiroyuki Namba
Profile
TitleOn Shortest Disjoint Paths by Computing Permanents
Abstract
Time16:00－17:00
SpeakerNaoki Marumo
Profile
Title
Abstract
Time17:00－18:00
SpeakerShinsaku Sakaue
Profile
Title
Abstract

Detail

DateSeptember 2, 2015 (Wednesday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:00－16:30
SpeakerSyuhei Denzumi
Profile
TitleAdvanced Hybrid DenseZDD and Recent Topics

Abstract
Time16:30－17:30
SpeakerYuni Iwamasa
Profile
Title
Abstract

Detail

DateAugust 31, 2015 (Monday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerISHII, Toshimasa
ProfileGraduate School of Economics and Business Administration,
Hokkaido University
Title
Abstract

Detail

DateAugust 19, 2015 (Wednesday)
Site
Time15:30－17:00
SpeakerJun Kawahara
ProfileGraduate School of Information Science, Nara Institute of Science and Technology, Japan.
Title
Abstract

Detail

DateJuly 10, 2015 (Friday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerHiroshi Sawada
ProfileNTT Service Evolution Laboratories
Title
Abstract

Detail

DateJuly 7, 2015 (Tuesday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerKimihito Ito
ProfileDivision of Bioinformatics Hokkaido University Research Center for Zoonosis Control
TitleCombining Population Genetics of Pathogens and Epidemiology of Infectious Diseases

AbstractThrough efforts combining mathematics, informatics, epidemiology, and biology, we are developing computational methods for the prediction and prevention of zoonotic outbreaks and epidemics. I this presentation, I will talk about following three topics:

1. Detecting changes in viral population size by population genetics

Early detection of ongoing outbreaks of influenza in non-natural host species is critical to control both avian and human influenza. Here, we propose the use of Tajima’s D as a tool to detect changes in viral population size. By analyzing 6,782 nucleotide sequences from 21 published research papers, we show that the mean values of Tajima’s D are different depending on the hosts and segments. The Tajima’s D values of internal influenza virus genes isolated from wild mallard samples are close to zero, meaning that the population of influenza A virus is stable. However, the Tajima’s D values from chicken and human samples are negative, which means the viral population is increasing. We anticipate this method can be applied to other pathogens to detect hidden outbreaks.

2. A large-scale analysis of interspecies transmission of zoonotic pathogens

We are developing a computational method that detects the transmission of pathogens between different host species. A total of 33,587 nucleotide sequences of the hemagglutinin and neuraminidase of avian and swine influenza viruses were downloaded from the internet. To clarify the interspecies transmission of influenza viruses between pigs and birds, we employed the reciprocal best BLAST hits algorithm. Those pairs sharing genes of 100% identity were regarded as the evidence of interspecies transmissions between avian and swine viruses. The method detected one hundred five possible interspecies transmissions between birds and pigs. Twelve results of them are consistent with the results from scientific papers that were published previously, suggesting that our method can correctly detect interspecies transmission. We anticipate that this method can be applied to detect the other zoonotic pathogens.

3. Predicting antigenic changes of influenza viruses through data assimilation

Human influenza A viruses undergo antigenic changes with gradual accumulation of amino acid substitutions on the hemagglutinin molecule. Antigenic mismatch between vaccine and epidemic strains often requires the replacement of influenza vaccine strains. To establish a practical method enabling us to predict the future direction of the viral evolution, we are developing a new prediction method based on a computational technique called data assimilation. Our aim here is to integrate actual observations of viral gene mutation into computer simulations, and infer current herd immunity and next mutations. To establish a practical method enabling us to predict amino acid substitutions on the hemagglutinin, we have constructed a mathematical model of viral population, infection, and host immunity. Based on the developed model, actual viral evolution observed in past 45 years was analyzed by particle filters. The timing when the dominant epidemic strains were replaced by other strains—as well as the future direction of the viral evolution—could be predicted by the method.

Detail

DateMay 15, 2015 (Friday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerDavid Avis
ProfileDepartment of Communications and Computer Engineering, School of Informatics, Kyoto University
TitlePolynomial size matching polytopes
AbstractI will begin by giving an introduction to extension complexity including
the exponential lower bound results of Fiorini et. al for the travelling
salesman problem and of Rothvoss for the matching problem. This latter
result proves the striking result that Edmonds' matching polytope is not
the projection of any polynomially sized polytope.

I will then describe a perfect matching polytope that is different from
Edmonds' polytope and explain the notion of a weak extended
formulation(WEF). Then I will show that this new polytope has a WEF of
polynomial size. This implies that perfect matchings in graphs can be
solved in polynomial time by linear programming over the WEF. We have
built a compiler that inputs an algorithm in pseudocode and outputs a
corresponding WEF.
This is joint work with David Bremner, Hans Tiwary and Osamu Watanabe.

Detail

DateMay 12, 2015 (Tuesday)
Site
Time15:30－17:00
SpeakerTakeru Inoue
ProfileNetwork Innovation Laboratories, NTT
Title
Abstract

Detail

DateApril 27, 2015 (Monday)
SiteC302 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerRyoma Sinya
ProfileShudo Lab., Tokyo Institute of Technology
Title
Abstract

Detail

DateFebruary 6, 2015 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－17:00
SpeakerMatias Korman
ProfileAssistant Professor at NII (Network Algorithm Group of the ERATO Kawarabayashi Large Graph Project)
TitleExtremal distances in Polygonal Domains
AbstractThe geodesic distance between two points p,q in a polygonal domain P is defined as the path of shortest length connecting the two points that contained in P. This distance is a proper metric, and as such the usual concepts of diameter (i.e., points of P that are furthest away) and center (i.e., point whose distance to its furthest away point is minimized) naturally extend. Although the definition of these traits is very simple, finding efficient algorithms that can compute these values has proven to be difficult. In this talk, we discuss the difficulties that lie in designing algorithms for computing the center and diameter for both simple domains and in the presence of holes.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateJanuary 30, 2015 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－16:30
SpeakerDavid Avis
ProfileDepartment of Communications and Computer Engineering, School of Informatics, Kyoto University
TitleA portable parallel implementation of the lrs vertex enumeration code
AbstractI will begin by describing the vertex enumeration problem and a method to solve it using reverse search with its implementation lrs. Then I will describe a parallel implementation of lrs that automatically exploits available hardware on multi-core computers and runs on a wide range of platforms.
The implementation makes use of a C++ wrapper that essentially uses the existing lrs code with only minor modifications. It makes use of the restart feature of reverse search that allows for independent subtree search and the fact that no communication is required between these searches. As such it can be readily adapted for use in other reverse search enumeration codes.
Joint work with Gary Roumanis (Microsoft)
Connected siteTokyo, Kansai, Kyushu University

Detail

DateJanuary 16, 2015 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:30
SpeakerHiroki Nakahara
ProfileEhime University
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateDecember 25, 2014 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerSeika Abe
ProfileNTT DATA Mathematical Systems Inc.
TitleTBD
AbstractTBD
Connected siteTokyo, Kansai, Kyushu University

Detail

DateDecember 19, 2014 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time14:00－15:20
SpeakerKazuhisa Makino
ProfileResearch Institute for Mathematical Sciences, Kyoto University
TitleTBD
AbstractTBD
Connected siteTokyo, Kansai, Kyushu Univ., Kyoto Univ.
Time15:20－16:30
SpeakerTakeaki Uno
ProfileNational Institute of Informatics
TitleTBD
AbstractTBD
Connected siteTokyo, Kansai, Kyushu Univ., Kyoto Univ.

Detail

DateDecember 5, 2014 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerShogo Takeuchi, Ryutaro Kurai
ProfileJST ERATO
TitleIntroduction to SDD (Sentential Decision Diagram) and its software package
AbstractTBD
Connected siteTokyo, Kansai, Kyushu University

Detail

DateNovember 14, 2014 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerYasuyuki Shirai
ProfileJST ERATO
TitleUnderstanding Changes in Customer Purchase Behavior: Study of Attenuation Model for Multiparametric Purchase Preferences
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateOctober 30, 2014 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerTakashi Washio
ProfileOsaka University / JST ERATO
TitleMass based estimation of data distribution and dissimilarity
Abstract

Detail

DateOctober 24, 2014 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:00
SpeakerKoji Tuda
ProfileUniversity of Tokyo / JST ERATO
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateOctober 3, 2014 (Friday)
SiteKansai Lab.
Time15:00－17:00
SpeakerIsao Tanaka
ProfileDept. of Materials Science and Engineering, Graduate School of Engineering, Kyoto University
TitleTBD
AbstractTBD
Connected siteSapporo, Tokyo

Detail

DateSeptember 3, 2014 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－16:30
SpeakerTaiji Suzuki
ProfileTokyo Institute of Technology
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateAugust 25, 2014 (Monday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerMichael Lampis
ProfileResearch Institute for Mathematical Sciences, Kyoto University
TitleParameterized Approximation Schemes using Graph Widths
AbstractA number of natural graph problems are known to be W-hard to solve exactly when parameterized by standard widths (treewidth or clique-width). At the same time, such problems are typically hard to approximate in polynomial time. In this talk we will present a natural randomized rounding technique that extends well-known ideas and can be used to obtain FPT approximation schemes for several such problems, evading both polynomial-time inapproximability and parameterized intractability bounds.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateAugust 19, 2014 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:00
SpeakerKazuyoshi Sasaki
ProfileMiraikan
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateAugust 5, 2014 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time14:00－15:30
SpeakerShohei Hido
ProfilePreferred Infrastructure
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateJuly 15, 2014 (Tuesday)
SiteHokkaido University, Grad. School of Information Science and Technology, 2F A22
Time15:30－17:30
SpeakerKitsuchart Pasupa
ProfileKing Mongkut's Institute of Technology Landkrabang
TitleCan relevance of images be inferred from eye movements?
AbstractSearching for images from a large collection is a difficult task for automated algorithms. Many current techniques rely on items which have been manually 'tagged' with descriptors. This situation is not ideal, as it is difficult to formulate the initial query, and navigate the large number of hits returned. In order to present relevant images to the user, many systems rely on an explicit feedback mechanism. A machine learning algorithm can be used to present a new set of relevant images to the user -- thus increasing hit rates. In this work, we use eye movements to assist a user when performing such a task, and ask this basic question: "Is it possible to replace or complement scarce explicit feedback with implicit feedback inferred from various sensors not specifically designed for the task?" In reasonably controlled setups, fairly simple eye movements‘features in conjunction with machine learning techniques are capable of judging the relevance of an image based on eye movements alone, without using any explicit feedback -- therefore potentially assisting the user in a task. We also combines image features together with implicit feedback from users' eye movements, using them to rank images in the database.
Connected sitenone

Detail

DateJuly 7, 2014 (Monday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time13:00－14:30
SpeakerLeonardo Bartoloni
ProfileUniversity of Pisa
TitleTime/space tradeoff on Discrete Optimization using approximated Binary Decision Diagrams
AbstractA wide range of discrete optimization problems can be solved using encoding the set of feasible solutions on a binary decision diagrams (BDD), with weights associated to arcs, and then performing a shortest path search on the BDD graph. In the worst case however the size of the BDD can be of exponential order in respect to the number of variables, making this method unsuitable for a lot of practical problems.
In this talk we will show how by merging or discarding nodes during the BDD construction it is possible to generate simpler diagrams which encode a superset or a subset of feasible solutions.
With those diagrams we can solve relaxed and restricted versions of the optimization problem using a limited memory space.
Finally we will show how these approximation can be used as branch elimination heuristics in a branch-and-bound search in order to solve the original optimization problem.
Connected siteTokyo, Kansai, NTT(x2)

Detail

DateJuly 4, 2014 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:30－18:00
SpeakerHiroshi Tohjo
ProfileNetwork Innovation Laboratories, NTT
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateMay 28, 2014 (Wednesday)
SiteERATO Osaka Office
Time15:30－17:30
SpeakerAtsushi Takizawa
ProfileOsaka City University
TitleTBA
AbstractTBA
Connected siteSapporo, Tokyo

Detail

DateMay 7, 2014 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:00
SpeakerNaohide Nagatsu
ProfileNTT BizLink, Inc.
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateApril 30, 2014 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerYutaka Nagahata
ProfileGraduate School of Life Science, Hokkaido University
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateApril 16, 2014 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerValia Mitsou
ProfileVisiting researcher at Kyoto University / The Graduate Center, City University of New York
TitleComputational complexity of games and puzzles, and, in particular of
the card game SET
AbstractIn this talk, we will analyze the algorithmic properties of one and two-player games people enjoy playing, such as Sudoku or Chess. Questions asked about puzzles and games in this context are of the following type: can we design efficient computer programs that play optimally given any opponent (for a two-player game), or solve any instance of the puzzle in question?
We will first give an overview of the field describing how to attack this question and what is generally known for games and puzzles in this context. Alongside, we will give some required background.
Then, we will focus on the game of SET, a card game where the objective is to form sets of cards that match in a certain sense using cards from a special deck. We will analyse single- and multi-round variations of this game and establish interesting connections with other classical computational problems. We will prove algorithmic and hardness results inthe classical and the parameterized sense.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateApril 11, 2014 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerHiroyuki Hanada
ProfileHokkaido University
TitleAverage-case linear-time similar substring searching by the q-gram distance
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateFebruary 26, 2014 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time10:30－12:00
SpeakerTakeru Inoue
ProfileNTT
TitleGraphillion update
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University
Time13:30－14:15
SpeakerNorihito Yasuda
ProfileJST ERATO
TitleCase studies in enumerating subtrees using Graphillion
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University
Time14:15－15:00
SpeakerMasakazu Nakamoto
ProfileJST ERATO
TitleEkillion as an Applications of Graphillion
Abstract
Connected siteTokyo, Kansai, Kyushu University
Time15:30－17:30
SpeakerYuichiro Yasui
ProfileChuo University / JST CREST
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateJanuary 24, 2014 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerHiroaki Iwashita
ProfileJST ERATO
TitleTITLE TBA
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateJanuary 17, 2014 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerTakahisa Toda
ProfileJST ERATO
TitleDualization of Boolean functions Using Ternary Decision Diagrams
AbstractTBA
Connected siteTokyo, Kansai, Kyoto Univ., Yamanashi Univ., Kobe Univ., NII, Kyushu University

Detail

DateJanuary 10, 2014 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerCharles Jordan
ProfileGraduate School of Information Science and Technology, Hokkaido University
TitleTowards Large-Scale Experimental Descriptive Complexity
AbstractImagine that we hope to find some program or function satisfying certain properties. This is a very common task in program synthesis, machine learning and AI, however it also arises in complexity theory. There we might search for reductions between particular problems or fast programs for certain problems.
Descriptive complexity theory is a logical approach to complexity theory that represents programs and functions as formulas in various logics, where expressive power corresponds to computational resources. In this presentation, we give an overview of recent experimental work in descriptive complexity. Experimental descriptive complexity uses computers to try to understand issues in complexity theory and uses techniques from descriptive complexity in other fields. We give an introduction to basic notions from descriptive complexity and describe several of these applications. In particular, we focus on recent work that seeks to utilize large-scale systems (supercomputers, clusters, coprocessors).
Connected siteTokyo, Kansai, Kyushu University

Detail

DateDecember 16, 2013 (Monday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerPhillippe Samer
ProfileLaboratory for Algorithms, Graduate School of Information Science and Technology, Hokkaido University
TitleA branch and cut algorithm for spanning trees under conflict constraints
AbstractLet $G(V,E)$ be a finite, undirected graph, and $C \subset E \times E$ a set of conflicting edge pairs. This talk overviews an approach for the exact solution of the NP-hard problem of determining a minimum spanning tree of $G$ including at most one of the edges from each pair in $C$.
In this work, we regard polyhedral representations of conflict-free edge subsets as stable sets in an auxiliary conflict graph $H(E,C)$. We present integer linear programming formulations including four classes of exponentially-many constraints: two of which correspond to classic polyhedral representations of spanning trees in $G$, and two for strengthening the intersection with relaxations of the polytope of stable sets in $H$ (with clique and odd-cycle inequalities).
We introduce and evaluate a preprocessing method and branch and cut algorithms. Encouraging results consistently improve those previously available in the literature. New feasibility and optimality certificates are provided, and stronger dual bounds are already obtained in the initial linear relaxation of the formulations, even for the hardest instances in the standard benchmark.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateDecember 10, 2013 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time13:00－14:30
SpeakerNorihito Yasuda
ProfileJST ERATO
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu University, Saitama Univ., Tokai Univ.

Detail

DateDecember 6, 2013 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:30
SpeakerNabila Abdessaied
ProfileInstitute of Computer Science, Group of Computer Architecture, University of Bremen, Germany
TitleUpper bounds for reversible circuits based on Young subgroups
AbstractWe call a logic function f : {0,1}^n -> {0,1}^n reversible if it is bijective, i.e. there exists a 1-to-1 mapping from the inputs to the outputs.
Synthesis is the problem of finding a sequence of reversible functions g1; g2; : : : ; gk such that f = g1*g2*...*gk where * denote a decomposition and each gi is from a gate set G.
G is called universal if such a sequence exists for all functions. Toffoli gates Tn are a universal gate set. We are interested in determining a tighter upper bounds on the number of Toffoli gates needed for synthesizing a reversible function. Both multiple controlled Toffoli gates and mixed polarity Toffoli gates have been considered for this purpose.
The calculation of the bounds is based on a synthesis approach based on Young subgroups that results in circuits using a more generalized gate library.
Starting from an upper bound for this library we derive new bounds which improve the existing one by around 77%.
Connected siteTokyo, Kansai, Iwate Univ.

Detail

DateDecember 3, 2013 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time13:00－14:30
SpeakerYoshihiko Suhara
ProfileNTT
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University, NII, Tokai Univ.

Detail

DateNovember 28, 2013 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:45－17:45
SpeakerWill Archer Arentz
ProfileExecutive Scientist and Vice Director of Rakuten Institute of Technology
TitleWhat is Rakuten Institute of Technology (RIT)?
AbstractAn informal introduction Rakuten Institute of Technology - who and what we are, as well as some of the things we work on. Some examples of research and development within distributed systems, algorithms, web services, financial services and O2O will be given.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateNovember 8, 2013 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－18:00
SpeakerYasuyuki Shirai
ProfileJST ERATO
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateNovember 1, 2013 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－18:00
SpeakerTakashi Washio
ProfileOsaka University / JST ERATO
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateOctober 28, 2013 (Monday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerYasuo Tabei
ProfileJST-PRESTO project
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Tsukuba, Kyoto Univ., Mr. Nakahara

Detail

DateOctober 25, 2013 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－18:00
SpeakerKoji Tuda
ProfileAIST / JST ERATO
TitleData Mining in Biological Science
AbstractTBA
Connected siteTokyo, Kyoto Univ., Tsukuba Univ.

Detail

DateOctober 8, 2013 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerFlorian Lonsing
ProfileKnowledge-Based Systems Group, Vienna University of Technology
TitleSearch-based QBF Solving: Basics, Recent Trends and Challenges
AbstractThe logic of quantified Boolean formulas (QBF) is an extension of propositional logic where the Boolean variables in a formula are existentially or universally quantified. The decision problem of QBF is PSPACE-complete, which allows to use QBF as a natural modelling language for PSPACE-complete problems from model checking and formal verification, for example. In this talk, we give an overview on search-based QBF solving which is based on recursive variable instantiation with backtracking. Search-based QBF solving is related to the DPLL algorithm for SAT which was published in its original form in 1962. The implementations of modern search-based QBF solvers, however, largely differ from the basic recursive approach. Typically, QBF solvers apply sophisticated rules to reduce the number of variables that must be assigned to solve a QBF. For example, the resolution rule for QBF allows solvers to deduce new clauses from a given QBF, which is called clause learning. The application of these rules in modern QBF solvers blurs the traditional picture of search-based QBF solving as a recursive algorithm. In addition to giving a broad overview, we report on recent trends and challenges in search-based QBF solving. Thereby, we focus on the implementation of our QBF solver DepQBF.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateOctober 1, 2013 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－17:00
SpeakerYutaka Nagahata
ProfileGraduate School of Life Science, Hokkaido University
TitleCombination explosion problem in extraction of hierarchical structure from steady state Markov chain
AbstractIn this talk, I will present clustering criteria for steady state Markov chain network (or generally speaking, network with edge weight) and combination explosion problem in its clustering algorithm. The problem has been discussed in study of spectral theory of graph (heuristics minimum conductance, random walk mixing time, diameter of graph) In the talk, I will address the importance of the problem especially in context of molecular science. I will also talk about relation to the minimum conductance cut in weighted network
Connected siteTokyo, Kansai, Kyushu University

Detail

DateSeptember 20, 2013 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－18:00
SpeakerRajeev Raman
ProfileUniversity of Leicester, UK
TitleEncodings for Range Selection and Top-k Queries
AbstractWe study the problem of encoding the positions the top-k elements of an array A[1..n] for a given parameter 1 \leq k \leq n. Specifically, for any i and j, we wish create a data structure that reports the positions of the largest k elements in A[i..j] in decreasing order, without accessing A at query time. This is a natural extension of the well-known encoding range-maxima query problem, where only the position of the maximum in A[i..j] is sought, and finds applications in document retrieval and ranking. We give (sometimes tight) upper and lower bounds for this problem and some variants thereof. (This work is joint with Roberto Grossi, John Iacono, Gonzalo Navarro, Rajeev Raman, Satti Srinivasa Rao and was presented at ESA 2013).
I will also talk about recent unpublished results for the encoding range selection problem, done jointly with a subset of the above co-authors.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateSeptember 18, 2013 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time11:00－12:30
SpeakerShin-ichiro Suzuki
ProfileMiraikan, Japan
TitleMuseum and the Web 2013 Report
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateSeptember 12, 2013 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:30
SpeakerYuko Sakurai
ProfileGraduate School and Faculty of Information Science and Electrical Engineering, Kyushu University
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateSeptember 3, 2013 (Tuesday)
SiteKansai Office
Time15:30－17:30
SpeakerNovi Quadrianto
ProfileNewton Fellow, Cambridge Machine Learning Group, University of Cambridge
TitleSimple and Efficient Bayesian Random Forest
AbstractRandom forests are one of the most successful ensemble methods in machine learning with state-of-the-art performance in many application domains. It works by averaging several predictions of de-correlated trees. Training phase of each decision tree involves finding data features to recursively partition the data space, and fitting a predictive model within each partition. All decision tree learning algorithms which are used in practise select features based on either entropy, or the Gini impurity criteria. In this talk, I will show a conceptually radical yet simple and efficient approach to generate a random forest based on Bayesian statistics. (Joint work with Zoubin Ghahramani.)
Connected siteHokkaido Univ., Tokyo Tech, Kyushu Univ.

Detail

DateAugust 29, 2013 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－17:00
SpeakerMartina Seidl
ProfileInstitute for Formal Models and Verification, Johannes Kepler University
TitleRecent Advancements in QBF Solving
AbstractQuantified Boolean formulas (QBF) provide a powerful framework for efficiently encoding application problems located in PSPACE including problems from formal verification, artificial intelligence, etc. The language of QBF extends propositional logic by universal and existential quantifiers over the propositional variables. The interest in QBF is raised by the vision of using QBF solvers as effectively as SAT solvers are used for problems in NP providing general, highly tuned reasoning engines.
In this talk, state-of-the-art techniques are presented which contribute towards making QBF technology ready for practical applications. Therefore, well established solving techniques are reviewed first including the QBF version of the famous DPLL decision procedure with learning which is implemented in most state-of-the-art QBF solvers. On this basis, it is shown how novel techniques like preprocessing, dual propagation, and certificate extraction are integrated into the solving workflow and how they affect the solving performance.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateAugust 28, 2013 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－17:00
SpeakerZhiyong Peng
ProfileWuhan University
TitleKey Management for Access Control in Trusted Cloud Storage
AbstractIn trusted cloud storage (TCS), for protecting the privacy of the sensitive outsourced cloud data, data owners locally encrypt their data before outsourcing. Through the secure management of the data keys, the selective access of outsourced data can be enforced in TCS scenarios. However, in TCS with multiple data owners, it remains a challenge to reduce users’ security risk and costs of key management as much as possible. In this talk, we propose a novel key management scheme based on global logical hierarchical graph (GLHG) for key derivation, which is used to enforce correctly the global authorization policies of all users. Our solution can achieve high efficiency by delegating the management of GLHG structure to cloud and adopting proxy re-encryption (PRE) technology. Additionally, we state the update policies for supporting dynamic access control. Finally, we show the benefits of our solution by experimentally evaluating quantitative criterions of key management.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateAugust 27, 2013 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－17:00
SpeakerTakuya Akiba
ProfileThe University of Tokyo
TitleTBA
Abstract
Connected siteTokyo, Kansai, Saitama Univ., Chuo Univ.

Detail

DateAugust 26, 2013 (Monday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerŁukasz Kaiser
Profile LIAFA in Paris
TitleExperimental Descriptive Complexity and Machine Learning
AbstractDescriptive complexity studies the logical means necessary to define a function or set of interest. In addition to standard complexity questions (about the time, space, or communication needed) and circuit complexity issues (such as the circuit size or depth necessary), descriptive complexity studies structural properties of definitions, such as uniformity and locality. I will introduce these notions and show that they are closely related to recent developments in machine learning. In particular, I will show that an extension of first-order logic with bit and successor relations and the majority quantifier, which has been widely studied in descriptive complexity, corresponds to a class of convolutional neural networks that have recently been applied in many domains. I will discuss how this relationship can be used both in experimental descriptive complexity and to gain insights into machine learning algorithms.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateAugust 8, 2013 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－17:00
SpeakerHiroki Nakahara
ProfileKagoshima University
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateAugust 1, 2013 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:30－18:00
SpeakerTakehiro Ito
ProfileTohoku University
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateJuly 16, 2013 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－16:30
SpeakerShinobu Nagayama
ProfileHiroshima City University
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateJuly 4, 2013 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time14:00－16:00
SpeakerTakanori Maehara
ProfileNII / JST ERATO
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateJuly 2, 2013 (Tuesday)
SiteOsaka Office
Time16:30－18:30
SpeakerMarco Cuturi
ProfileAssociate Professor, Graduate School of Informatics, Kyoto University
TitleLightspeed Optimal Transportation Distances for Histograms
AbstractOptimal transportation distances are a fundamental family of parameterized distances for histograms. Despite their appealing theoretical properties, excellent performance and intuitive formulation, their computation involves the resolution of a linear program whose cost is prohibitive whenever the histograms' dimension exceeds a few hundreds. We will try to solve this challenge in this talk using two different sets of tools:

- Kernel methods: we prove that the classical transportation distance can be turned into a positive definite kernel by considering the generating function of the transportation polytope. Since this quantity is computationally intractable we study random sampling approaches by sampling vertices of the transportation polytope.The resulting kernel has a linear time complexity and performs slightly
worse than regular EMD in our experimental studies.

- Regularization: we smooth the classical optimal transportation problem with an entropic regularization term, and show that the resulting optimum is also a distance which can be computed through Sinkhorn-Knopp's matrix scaling algorithm. Through this trick, we avoid the computation of an LP and we obtain a distance which is up to 1.000.000 x faster to compute than classical EMD and can be parallelized on GPGPUs. This dramatic speed-up opens the door for large-scale applications of optimal transportation distances. We also report improved performance on the MNIST dataset.
Connected siteSapporo, Tokyo, Kyushu Univ.

Detail

DateMay 29, 2013 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time13:30－15:30
SpeakerNorihito Yasuda
ProfileNTT
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateMay 9, 2013 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerFlorian Horn
ProfileCNRS researcher at LIAFA
TitleStochastic Games in Verification
AbstractGames are a classical tool for the synthesis of controllers in reactive systems. In this setting, the system and its evolution are represented by a graph (the arena), on which two players (Adam and Eve) move a token for an infinity of steps. The winner of a play is decided by a subset of the infinite sequences of steps.
In this talk, I will introduce the important concepts of game theory in this field, and present some of my results about the usual problems the we consider : How to decide the winner of a given game? What kind of strategies do the players need to achieve their objectives? How much memory do these strategies require ?
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateApril 24, 2013 (Wednesday)
SiteOsaka Office
Time15:00－17:00
SpeakerMichael E. Houle
ProfileVisiting Professor, National Institute of Informatics
TitleIntrinsic Dimensionality and Discriminability of Data
AbstractIn this presentation, we examine the uses of a measure of intrinsic dimensionality, the "generalized expansion dimension", in a variety of applications in search and data mining. The original formulation of expansion dimension was formulated in 2002 by Karger and Ruhl for the theoretical analysis of search indices. However, generalized versions of this measure turn out to be quite useful not only theoretically, but also as the basis of runtime testing. We will summarize two results that were presented at IEEE ICDM 2012, one on practical methods for estimation of intrinsic dimensionality (presentation at a workshop associated with ICDM 2012), and the other on an application of intrinsic dimensional testing for search using lower-bounding distance functions (a full-paper presentation at the main conference).
A brief survey will also be provided of other papers presented at ICDM 2012 that may be of interest to members of the ERATO Minato project. The talk will conclude with a report on ongoing work on a continuous variant of intrinsic dimension for arbitrary data distributions.
Connected siteSapporo, Tokyo, Kyushu Univ.

Detail

DateApril 12, 2013 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerKazuki Yoshizoe
ProfileResearch, JST ERATO
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateMarch 26, 2013 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－16:30
SpeakerMasanori Sugimoto
ProfileProfessor, Hokkaido University
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Time16:30－18:00
SpeakerTetsuya Yoshida
ProfileHokkaido University
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateMarch 22, 2013 (Friday)
SiteTokyo Satellite Lab (Tokyo Tech.)
Time15:00－17:00
SpeakerMasakazu Iahihata
ProfileTokyo Institute of Technology
TitleProposionalized Probability Computation and Learning on Binary Decision Diagrams
AbstractTBA
Connected siteSapporo, Osaka, Kyushu Univ.

Detail

DateMarch 21, 2013 (Thursday)
SiteTokyo Satellite Lab (Tokyo Tech.)
Time15:00－16:30
SpeakerMartin Müller
Profile University of Alberta
TitleMove Quality in Monte Carlo Simulation: A Case Study using the Fuego Go Program
AbstractDespite half a dozen years of intense research, designing effective simulation policies for Monte Carlo Tree search in Go is still considered something of a black art, and driven largely by trial and error. Important ideas that have evolved include pattern- and tactics-based playouts, simulation balancing, and several schemes to dynamically modify simulation policies online. In this study, we take an in-depth look at what happens when the Go program Fuego runs its playouts. We develop several methods to evaluate the quality of moves played in these simulations, and we evaluate the contribution of the different components of Fuego's playout policy. We study the distribution of both the number of blunders, or result-changing moves, and the absolute loss - in terms of number of points - for many variations of the Fuego playout policy. We use this study to identify an improvement to the Fuego default policy.
Connected siteSapporo, Osaka, Kyushu Univ.

Detail

DateMarch 11, 2013 (Monday)
SiteOsaka Office
Time16:00－18:00
SpeakerToshiki Saitoh
ProfileGraduate School of Engineering Faculty of Engineering, Kobe University
TitleTBA
AbstractTBA
Connected siteSapporo, Tokyo, Kyushu Univ.

Detail

DateMarch 9, 2013 (Saturday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time10:30－12:00
SpeakerHiroaki Iwashita, Yoshio Nakawaza
ProfileJST ERATO, Free Programmer
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu University

Detail

DateJanuary 31, 2013 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:30
SpeakerShin Miyakawa
ProfileNTT Communications
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateJanuary 28, 2013 (Monday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time13:00－16:00
SpeakerRobert Wille and Mathias Soeken
ProfileUniversity of Bremen
TitleSynthesis of Reversible Circuits using Decision Diagrams: Background,Recent Accomplishments, and Open Challenges
AbstractDue to its promising applications in domains such as quantum computation and low-power design, synthesis of reversible circuits has become an intensely studied topic. However, many synthesis methods are limited by non-scalable function representations such as truth tables. As an alternative, synthesis exploiting graph-based representations has been suggested. The underlying structure is a decision diagram (DD) that may vary regarding reduction methods, decomposition rules, or ordering restrictions.
In this talk, we are proving a comprehensive overview on this particular research area. First, we review the background on reversible circuits and their applications as an emerging technology followed by a summary of the recent accomplishments in the synthesis of these circuits using decision diagram. With this as basis, the second part of the talk outlines open research problems and provides possible solutions as well as recent work in progress on how to solve these problem.
Connected siteTokyo, Kansai, Kyushu University

Detail

DateJanuary 15, 2013 (Tuesday)
SiteOsaka Office
Time16:30－18:00
SpeakerRyo Yoshinaka
ProfileGraduate School of Informatics, Kyoto University
TitleTBA
AbstractTBA
Connected siteSapporo, Tokyo, Kyushu Univ.

Detail

DateJanuary 11, 2013 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:30
SpeakerShin-ichi Minato
ProfileHokkaido University
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai, Kyushu Univ., Kyoto Univ.

Detail

DateDecember 12, 2012 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:00－15:30
SpeakerToshio endo
ProfileTokyo Institute of Technology, JST CREST
TitleTBA
Abstract
Connected siteTokyo, Osaka, Kyushu Univ.
Time15:30－16:00
SpeakerHitoshi Sato
ProfileTokyo Institute of Technology, JST CREST
TitleTBA
AbstractTBA
Connected siteTokyo, Osaka, Kyushu Univ.
Time16:00－16:45
SpeakerYuichiro Yasui
ProfileChuo University, JST CREST
TitleTBA
Abstract
Connected siteTokyo, Osaka, Kyushu Univ.
Time16:45－17:30
SpeakerKatsuki Fujisawa
ProfileChuo University, JST CREST
TitleTBA
AbstractTBA
Connected siteTokyo, Osaka, Kyushu Univ.

Detail

DateDecember 6, 2012 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－18:00
SpeakerShoji Kasahara
ProfileGraduate School of Information Science, Nara Institute of Science and Technology
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateDecember 5, 2012 (Wednesday)
SiteTokyo Satellite Lab (Tokyo Tech.)
Time13:30－15:30
SpeakerRajeev Raman
ProfileUniversity of Leicester, UK
TitleDynamizing Succinct Tree Representations
AbstractWe consider succinct, or space-efficient, representations of ordinal trees. Representations exist that take 2n + o(n) bits to represent a static n-node ordinal tree - close to the information-theoretic minimum - and support navigational operations in O(1) time on a RAM model; and some implementations have good practical performance. The situation is different for dynamic ordinal trees. Although there is theoretical work on succinct dynamic ordinal trees, there is little work on the practical performance of these data structures. Motivated by applications to representing XML documents, in this paper, we report on a preliminary study on dynamic succinct data structures. Our implementation is based on representing the tree structure as a sequence of balanced parentheses, with navigation done using the min-max tree of Sadakane and Navarro (SODA '10). Our implementation shows promising performance for update and navigation, and our findings highlight two issues that we believe will be important to future implementations: the difference between the finger model of (say) Farzan and Munro (ICALP ’09) and the parenthesis model of Sadakane and Navarro, and the choice of the balanced tree used to represent the min-max tree.
Connected siteSapporo, Osaka, Kyushu Univ.

Detail

DateDecember 5, 2012 (Wednesday)
SiteTokyo Satellite Lab (Tokyo Tech.)
Time15:30－17:30
SpeakerRoberto Grossi
ProfileUniversity of Pisa, Italy
TitleEmpowering Succinct Data Structures for Strings and Sequences
AbstractThe talk describes some problems that arise in industrial applications and how succinct and compressed data structures for strings and sequences can help using little space. The first part of the talk is devoted to semi-indexing semi-structured data, e.g. query log data in JSON format. The second part involves prefix searching on a set of strings which is helpful e.g. in search engines, and describes some succinct data structures for storing a set of strings. The third part of the talk considers a sequence of strings (not to be confused with a set of strings) and introduces the wavelet trie for storing textual data (e.g. URLs) in time order and supporting extra functionalities that exploit this order. The discussed results are joint works with Giuseppe Ottaviano presented at ACM CIKM 2011, ACM-SIAM ALENEX 2012, and ACM PODS 2012.
Connected siteSapporo, Osaka, Kyushu Univ.

Detail

DateNovember 22, 2012 (Thursday)
SiteOsaka Office
Time13:00－15:00
SpeakerAlexandre Termier
ProfileAssitant Prof., LIG (Laboratoire d'Informatique de Grenoble), Universite Joseph Fourier, France
TitleParaMiner: generic and parallel pattern mining scaling to real-world data
AbstractIn this talk, we present the PhD works of Benjamin Negrevergne, conduced at the University of Grenoble. The goal of these works is to propose an efficient parallel algorithm than can solve a broad range of pattern mining problems. In this regard, we exploit the results on polynomial-delay / polynomial space enumeration of Arimura and Uno [SDM'09] and of Boley et al. [Th.Comp.Sci.'10] on strongly accessible set systems. We pair these works with a novel dataset reduction technique, EL-reduction, that exploits some characteristics of the enumeration strategy for reducing the dataset. We show that this combination allows to mine real world datasets with a generic algorithm, with performances comparable with specialized algorithms, for a broad range of pattern mining problems. We will also present some recent developments on the problems of doing a dataset reduction in parallel on a multicore processor, and a solution we found for alleviating some of these problems and improving parallel scaling.

(*) Benjamin Negreverge's PhD was supervised by Alexandre Termier, Marie-Christine Rousset and Jean-Francois Mehaut.
Connected siteSapporo, Tokyo, Kyushu Univ.

Detail

DateNovember 19, 2012 (Monday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerNorie Fu
ProfileThe University of Tokyo
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateNovember 14, 2012 (Wednesday)
SiteTokyo Satellite Lab (Tokyo Tech.)
Time15:30－17:30
SpeakerKohei Hayashi
ProfileThe University of Tokyo
TitleTBA
Abstract
Connected siteSapporo, Osaka, Kyushu Univ.

Detail

DateNovember 12, 2012 (Monday)
SiteTokyo Satellite Lab (Tokyo Tech.)
Time15:00－17:00
SpeakerS.R.K. Branavan
ProfileChief Scientist, Natural Language Processing at Millennium Information Technologies, Colombo, Sri Lanka
TitleGrounding Linguistic Analysis in Control Applications
AbstractNatural languages are the medium in which the majority of humanity's collective knowledge is recorded and communicated. If machines were able to automatically access and leverage this information, they could effectively perform many tasks that are currently considered to be computationally intractable, thus requiring human involvement. Today, the only way to infuse human knowledge into computational algorithms is to have humans in the loop - i.e., to manually encode the knowledge into heuristics, through annotations, or directly into the model
structure itself.
Our ultimate goal is to automate this process, so that machines can access required knowledge directly from text. One path to this goal is to perform a semantic interpretation of text by grounding the textual information in the objects, actions and dynamics of the physical world. From a linguistic viewpoint, this grounding of language in control applications presents a very natural notion of language semantics. I.e., by allowing us to define semantics with respect to the control application, and avoid imposing subjective human notions of correctness.
In this talk, I will explore two aspects of the connection between language and control applications: first, how the semantic analysis of language can be driven by control performance; and second, how information from text can be leveraged to improve performance in complex control systems. These two aspects are in fact complementary, and language analysis is central to them both. Addressing these aspects jointly is particularly challenging. However, as our empirical results show, connecting the semantic analysis of language to control applications allow us to leverage the synergy between the two to simultaneously learn both language analysis and application control with little or no prior knowledge.
Connected siteSapporo, Osaka, Kyushu Univ.

Detail

DateNovember 9, 2012 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:30
SpeakerYasuyuki Shirai
ProfileJST ERATO
TitleSet-Similarity Joins
AbstractIn recent years, researches on similarity joins have significantly attentions for database error correction, data cleaning and so on. In order to handle with database whose entry count is over several million, efficient algorithms can be developed for set similarity joins. In this talk, I will briefly introduce current research activities on these areas, such as filtering based approach and trie-based approach. We also introduce ZDD based system, called ZDD-Join, which has been developed in our project.
Connected siteTokyo, Osaka, Kyushu Univ.

Detail

DateNovember 6, 2012 (Tuesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerTakahisa Toda
ProfileMinato Discrete Structure Manipulation System Project
TitleOn Separating Convex Points with Lines
Abstract
Connected siteTokyo, Osaka, Kyushu Univ.

Detail

DateNovember 2, 2012 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:30
SpeakerTakashi Washio
ProfileOsaka University / JST ERATO
TitleGlobal Maximization of Submodular Function and Its application to Machine Learning
AbstractFeature selection is a highly important task to enhance generality and transparency of multi-variate prediction models including classifiers and regression models in machine learning and statistics. It also contributes to identify feature variables (or factors) dependent on the objective prediction variable in the generation process of a given data set. Particularly, the globally optimum feature set providing the minimum prediction error of the model should be explored for the later purpose under the assumption that the prediction model captures the given data generation process. The model prediction error is known to be a submodular function of a feature subset where improvement of the error is less when the size of the selected feature subset gets larger, and thus the task to derive the globally optimum feature set is a submodular optimization problem which is NP-hard.
Based on this background, my talk addresses the development of an efficient algorithm for the global submodular maximization and its application to the optimum feature selection .
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateOctober 26, 2012 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time16:00－17:30
SpeakerKoji Tsuda
ProfileAIST / JST ERATO
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateOctober 3, 2012 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerLukasz Kaiser
ProfileLIAFA, CNRS & Universite Paris 7
TitleLearning and Playing Board Games using Descriptive Complexity
AbstractImagine that a few moves of an unknown game are demonstrated to you, as well as some winning and losing positions, and your task is to guess the rules of the game and then to develop a method to play it competitively. While this task might seem to be unrelated to mathematical logic and even a bit imprecise, I will argue that using descriptive complexity one can give a convincing precise definition. To give an algorithmic solution, I will introduce a few techniques from finite model theory, such as the Weisfeiler-Lehman algorithm, the guarded fragment and Ehrenfeucht-Fraisse games. Finally, I will demonstrate the resulting program on a few simple games and discuss other possible applications of descriptive complexity in program learning.
Connected siteTokyo, Osaka, Kyushu Univ.

Detail

DateSeptember 14, 2012 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerHidetomo Nabeshima
ProfileYamanashi University
TitleThe principles of the high performance SAT solvers
Abstract

Detail

DateSeptember 13, 2012 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerSeiichiro Tani
ProfileNTT
TitleTBA
Abstract

Detail

DateSeptember 12, 2012 (Wednesday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time13:30－15:30
SpeakerProf. Rusins Freivalds
ProfileInstitute of Mathematics and Computer Science, University of Latvia, Riga, Latvia
TitleUltrametric automata and Turing machines
AbstractWe introduce a notion of ultrametric automata and Turing machines using p-adic numbers to describe random branching of the process of computation. These automata have properties similar to the properties of probabilistic automata but complexity of probabilistic automata and complexity of ultrametric automata can differ very much.
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateSeptember 7, 2012 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerTakashi Horiyama
ProfileSaitama University
TitleTBA
Abstract
Connected siteTokyo, Osaka, Kyushu

Detail

DateAugust 10, 2012 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time10:30－12:00
SpeakerYoshihiro Yamanishi
ProfileKyushu University
TitleMachine learning methods to infer drug-target interaction networks
AbstractMost drugs are small chemical compounds which interfere with the
behavior of their target proteins. Therefore, the genome-wide
detection of drug-target interactions or more generally
compound-protein interactions from heterogeneous biological data is a
key area in chemogenomics and genomic drug discovery toward
identification of new drug leads and therapeutic targets for known
diseases such as cancers.
In this research, we investigate the correlation between the chemical
space of drugs (e.g., chemical structures, fragments), the genomic
space of genes or proteins (e.g., sequences, domains, functional
sites), and the pharma- cological space of phenotypic effects (e.g.,
efficacy, side-effects, adverse drug reactions) in terms of the
topology of drug-target interaction networks. We then develop a new
method to predict unknown drug-target interactions from chemical,
genomic, and pharmacological data on a large scale. The prediction is
performed based on the state-of-art machine learning technology,
assuming that drug molecules with similar chemical structures and
similar phenotypic effects are likely to interact with similar target
proteins. The originality of this research lies in the formalization
of the drug-target interaction inference as a supervised learning
problem for a bipartite graph, the lack of need for 3D structure
information of the target proteins, and in the integration of
chemogenomic approach and pharmacogenomic approach in a unified
framework. Our comprehensively predicted drug-target interaction
networks enable us to suggest many potential drug-target interactions
and compound-protein interactions.
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateAugust 3, 2012 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time10:30－12:00
SpeakerShin Aida
ProfileToyohashi University of Technology
TitleTBA
Abstract
Connected siteTokyo, Osaka, Kyushu Univ.

Detail

DateJuly 6, 2012 (Friday)
SiteOsaka Office
Time15:30－17:30
SpeakerMutsunori Banbara
ProfileKobe University
TitleTBA
Abstract
Connected siteSapporo, Tokyo, Kyushu Univ.

Detail

DateJune 8, 2012 (Friday)
SiteERATO C304 Seminar Room
Time15:30－17:30
SpeakerIchigaku Takigawa
ProfileHokkaido University
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu University

Detail

DateMay 30, 2012 (Wednesday)
SiteERATO C304 Seminar Room

Time14:30－16:00
SpeakerTakeshi Yamamuro
ProfileNTT
TitleCompression and Searches with Modern Processors (Practice and Implementation )
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateMay 30, 2012 (Wednesday)
SiteERATO C304 Seminar Room
Time16:30－17:30
SpeakerMakoto Onizuka
Profile
Title
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateMay 24, 2012 (Thursday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerShogo Takeuchi
ProfileJST-ERATO MINATO Discrete Structure Manipulation System Project
Title
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateApril 20, 2012 (Friday)
SiteC304 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerHiroshi Aoki
ProfileRitsumeikan University
TitleTBA
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateApril 13, 2012 (Friday)
SiteC306 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerKensuke Fukuda
ProfileNational Institute of Informatics
Title
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateMarch 30, 2012 (Friday)
SiteC306 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerShi-ichiro Suzuki
ProfileNational Museum of Emerging Science and Innovation
TitleTBA
Abstract
Connected siteKansai, Tokyo, Kyushu

Detail

DateFebruary 24, 2012 (Friday)
SiteTokyo Satellite Lab (Tokyo Tech.)
Time15:30－17:30
SpeakerKazuki Yoshizoe
ProfileThe University of Tokyo
TitleTBA
Abstract
Connected siteSapporo, Tokyo

Detail

DateFebruary 20, 2012 (Monday)
SiteC306 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerMasakazu Iahihata
ProfileTokyo Institute of Technology
TitleBayesian networks and BDDs
Abstract

Detail

DateFebruary 10, 2012 (Friday)
SiteC306 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerCharles Jordan
ProfileGraduate School of Information Science and Technology, Hokkaido University
TitleDescriptive Programming
AbstractImagine we have some problem that we hope to solve, and so we write a program to compute the answer. Programming is time-consuming and error-prone, and developing optimized programs for hard problems is especially difficult.
In many cases, we don't care exactly how the answer is computed -- we just want to answer our problem. Instead of specifying exactly how the computation will proceed, couldn't we just pose our problem in some formal language and get an answer?
I'll describe recent joint work on a descriptive programming environment (DE). DE implements the objects of descriptive complexity theory in a programming environment, allowing users to automatically generate counter-examples to conjectures, examine reductions between problems, and rapidly develop reasonably-efficient programs for various problems.
DE is intended to be useful for researchers and students (we use it!), but it could be used more widely. We'll discuss related work and future plans, while focusing on examples where it could be useful. We encourage everyone to experiment with DE and would love to hear of any ideas, insights or improvements.
Connected siteTokyo, Kansai

Detail

DateJanuary 20, 2012 (Friday)
SiteTokyo Satellite Lab
Time15:30－17:30
SpeakerTetsuo Shibuya
ProfileHuman Genome Center, University of Tokyo
Title
Abstract
Connected siteSapporo, Kansai, Kyushu

Detail

DateDecember 26, 2011 (Monday)
SiteC306 ERATO Seminar Room, Hokkaido University
Time15:30－17:30
SpeakerTakahisa Toda
ProfileKyoto University
TitleTBA
Abstract
Connected siteTokyo, Kansai

Detail

DateDecember 19, 2011 (Monday)
SiteHokkaido University Eng.C306 ERATO Seminar Room
Time10:30－12:00
SpeakerKeisuke Hotta
ProfileBunkyo University
TitleTBA
Abstract
Connected siteTokyo, Kansai

Detail

DateNovember 25, 2011 (Friday)
SiteERATO Seminar Room
Time16:00－18:00
SpeakerTakashi Washio
Profile
Title
Abstract

Detail

DateNovember 24, 2011 (Thursday)
SiteERATO Seminar Room
Time16:00－18:00
SpeakerYasuyuki Shirai
Profile
TitleIntroduction to data mining application using large-scale data
Abstract

Detail

DateNovember 14, 2011 (Monday)
SiteERATO C304 Seminar Room
Time15:30－17:30
SpeakerShingo Yoshizawa
ProfileGraduate School of Information Science and Technology, Hokkaido University
Title
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateOctober 27, 2011 (Thursday)
SiteERATO C304 Seminar Room
Time15:30－17:00
SpeakerTetsuo Asano
ProfileJAIST
TitleDesigning Algorithms with Limited Work Space
AbstractRecent progress in computer systems has provided programmers with virtually unlimited amount of work storage for their programs. This leads to space-inefficient programs that use too much storage and become too slow if sufficiently large memory is not available. Thus, I believe that space-efficient algorithms or memory-constrained algorithms deserve more attention.

Constant-work-space algorithms have been extensively studied under a different name, log-space algorithms. Input data are given on a read-only array of $n$ elements, each having O(log n) bits, and work space is limited to O(log n) bits, in other words, a constant number of pointers and counters, each of O(log n) bits. This memory constraint in the log-space algorithms may be too severe for practice applications. For problems related to an image with n pixels, for example, it is quite reasonable to use work space of the order of square root of n, which amounts to a constant number of rows and columns.

I will start my talk with a simple algorithm for detecting a cycle in a graph using only some constant amount of work space (more exactly, O(log n) bits in total) and then its applications. Then, I will introduce some paradigms for designing such memory-constrained algorithms and their applications to interesting problems including those in computational geometry and computer vision.
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateOctober 14, 2011 (Friday)
SiteERATO Seminar Room
Time16:00－18:00
SpeakerKoji Tuda
Profile
TitleUltra-fast algorithms for all pairs similarity search and their application to biological data.
Abstract

Detail

DateOctober 12, 2011 (Wednesday)
Site
Time15:30－17:30
SpeakerEtsuji Tomita
Profile
Title
Abstract
Connected siteTokyo, Kansai, Saitama University

Detail

DateSeptember 30, 2011 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerEishi Chiba
ProfileHosei University
Title
AbstractUndecided
Connected siteTokyo, Kansai

Detail

DateSeptember 14, 2011 (Wednesday)
SiteERATO C304 Seminar Room
Time15:00－17:00
SpeakerHirokazu Anai
ProfileFujitsu Laboratories / Kyushu University
Title
Abstract

Detail

DateAugust 30, 2011 (Tuesday)
SiteERATO C304 Seminar Room
Time10:30－12:00
SpeakerJun Sakuma
ProfileTsukuba University
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateAugust 29, 2011 (Monday)
SiteERATO C304 Seminar Room
Time15:30－18:00
SpeakerTakashi Horiyama
ProfileSaitama University
Title
Abstract

Detail

DateAugust 26, 2011 (Friday)
SiteERATO C304 Seminar Room
Time15:30－17:30
SpeakerNaoki Kato
ProfileGraduate School of Engineering, Kyoto University
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateAugust 25, 2011 (Thursday)
SiteERATO C304 Seminar Room
Time10:30－12:00
SpeakerHiroki Nakahara
ProfileKyushu Institute of Technology
TitleTBD
Abstract
Connected siteTokyo, Kansai

Detail

DateAugust 25, 2011 (Thursday)
SiteERATO C304 Seminar Room
Time15:30－17:30
SpeakerAkihiro Kishimoto
ProfileTokyo Institute of Technology
TitleDepth-First Proof-Number Search in the Presence of Repetitions
AbstractDepth-first proof-number search (df-pn) is a powerful AND/OR tree search algorithm proven to be effective in solving positions in games. However, df-pn has a notorious problem of infinite loops when applied to domains with repetitions. Df-pn(r) practically cures it by ignoring proof and disproof numbers that may lead to infinite loops.

In this talk, I point out that df-pn(r) has a serious issue of underestimating proof and disproof numbers, while it also suffers from the overestimation problem occurring in directed acyclic graph. I present two practical solutions to these problems. While bypassing infinite loops, the threshold controlling algorithm (TCA) solves the underestimation problem by increasing the thresholds of df-pn. The source node detection algorithm (SNDA) detects the cause of overestimation and modifies the computation of proof and disproof numbers. Both TCA and SNDA are implemented on top of df-pn to solve tsume-shogi (checkmating problem in Japanese chess). Results show that df-pn with TCA and SNDA is far superior to df-pn(r). Our tsume-shogi solver is able to solve several difficult positions previously unsolved by any other solvers.
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateAugust 24, 2011 (Wednesday)
SiteERATO C304 Seminar Room
Time15:00－18:00
SpeakerKunihiko Sadakane
ProfileNational Institute of Informatics
TitleTBA
AbstractTBA
Connected siteTokyo, Kansai

Detail

DateAugust 23, 2011 (Tuesday)
SiteERATO C304 Seminar Room
Time10:30－12:00
SpeakerMasashi Sugiyama
ProfileTokyo Institute of Technology
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateAugust 23, 2011 (Tuesday)
SiteERATO C304 Seminar Room
Time15:30－17:30
SpeakerHiroo Ito
ProfileKyoto University
TitleTBA
AbstractTBA

Detail

DateAugust 9, 2011 (Tuesday)
SiteERATO C304 Seminar Room
Time15:30－17:30
SpeakerTakeaki Uno
ProfileNational Institute of Informatics
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateJuly 29, 2011 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerSeigo Takeuchi
ProfileThe University of Tokyo, Graduate School of Arts and Science
TitleTBA
Abstract(TBA)
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateJuly 15, 2011 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerMasaaki Nishino
Profile
Title
Abstract
Connected siteTokyo, Kansai, Saitama University

Detail

DateJuly 8, 2011 (Friday)
SiteERATO Seminar Room
Time15:30－18:00
SpeakerKazutaka Matsuzaki
ProfileMitsubishi Research Institute,Inc.
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateJuly 1, 2011 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerShin-ichi Minato
Profile Professor at Hokkaido University and Research Director of JST ERATO Minato Discrete Structure Manipulation System Project
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateJune 17, 2011 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerHiroaki Iwashita
ProfileERATO Minato Discrete Structure Manipulation System Project,
Researcher
TitleGraph Path Enumeration by ZDD-Mate Method
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateMay 24, 2011 (Tuesday)
SiteERATO seminar room
Time15:30－16:45
SpeakerProf. Neil Immerman
ProfileProfessor, University of Massachusetts Amherst
Title"P versus NP: Approaches, Rebuttals, and Does It Matter?"
AbstractNP contains all those problems that we could efficiently solve if "guessing were free", i.e., all those problems whose solutions we could recognize as correct if they were given to us in an appropriate format and context. We've known since 1971 that a large class of important combinatorial problems including boolean satisfiablity (SAT) are NP complete, i.e, hardest problems in NP.
If any of these had a solution in P, then so would all other problems in NP.

How hard is SAT, or any other NP-complete problem? It is well believed that P is not equal to NP and that SAT requires exponential time on some hard instances. However, SAT solvers are now in practical use as general problem solvers, currently working even on instances with a million variables.

I will explain some of these issues, laying out what we do know, what we do not know, and what we hope to understand once the P versus NP question, together with many similar but less famous open questions, are finally resolved. At the same time, this talk will present a survey of descriptive complexity -- the logical approach to complexity that I have been pursuing for years.
Connected siteTokyo, Kansai

Detail

DateMay 20, 2011 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerJun Kawahara
Profile
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateApril 21, 2011 (Thursday)
SiteERATO C304 Seminar Room
Time15:30－17:30
SpeakerShin-ichi Minato
ProfileProfessor at Hokkaido University and Research Director at JST ERATO Minato Project
Title
Abstract
Connected siteTokyo, Kansai, Kyushu Univ.

Detail

DateApril 15, 2011 (Friday)
SiteKansai Satellite Lab.
Time15:30－17:30
SpeakerShunji Umetani
Profile
Title
Abstract
Connected siteSapporo, Tokyo, Kyushu

Detail

DateApril 8, 2011 (Friday)
SiteKansai Satellite Lab.
Time15:30－17:30
SpeakerMichael Houle
ProfileNational Institute of Informatics
TitleIntrinsic Dimensionality and its Applications to Databases and Data Mining
AbstractIn recent years, important applications in the areas of machine learning, data mining, and bioinformatics have been developed that rely on the search for similarities among data structures as a fundamental operation. Typically, the computation of structural similarity is performed using a computationally-expensive exhaustive search over all possibilities. Effective indexing methods for structural similarity search, together with appropriate measures of inter-structure similarity, could potentially have a significant impact on the overall performance of such applications.

In this presentation, we propose a general framework within which the effectiveness of similarity search and other data operations can be assessed, independently of the nature of the data being considered. Within this framework, we discuss how a form of intrinsic dimensionality in the neighborhood of a test or query item, the "stereological dimension", can be assessed using two measurements of the distance and ranks of items within the neighborhood. Examples will be shown of the use of stereological dimension in the design and performance analysis of a variety of applications, such as similarity search, aggregation of search results, and anomaly detection (a paper describing the latter application received the Best Research Paper Award" at IEEE ICDM 2010). The presentation will conclude with a discussion of the potential of this framework for the development of effective applications based on structural similarity search.
Connected siteSapporo, Tokyo

Detail

DateMarch 23, 2011 (Wednesday)
SiteERATO Seminar Room
Time15:30－17:00
SpeakerShin-ichi Minato
ProfileProfessor at Hokkaido University and Research Director of JST ERATO Project
Title
Abstract
Connected siteTokyo, Kansai, Saitama University

Detail

DateMarch 22, 2011 (Tuesday)
SiteHokkaido University Eng.C306 ERATO Seminar Room
Time15:30－17:00
SpeakerShinichi Minato
ProfileProfessor at Hokkaido University and Research Director of JST ERATO Minato Project
Title
Abstract
Connected siteTokyo, Kansai, Saitama University

Detail

DateMarch 7, 2011 (Monday)
SiteERATO Seminar Room
Time15:30－17:30
Speaker Dr. Benjamin Rossman
ProfileDepartment of Electrical Engineering and Computer Science, MIT
TitleAverage-case complexity of detecting cliques
AbstractWe investigate the average-case complexity of the k-CLIQUE problem on random graphs with an appropriate density of edges. Our results are lower bounds of n^{k/4} for two well-studied classes of circuits: bounded-depth circuits and monotone circuits. Besides being the first lower bounds for k-CLIQUE in the average-case (and moreover essentially tight), these results lead to a new "size ierarchy theorem" for AC^0 and settle a longstanding open question in finite model theory.

Detail

DateFebruary 28, 2011 (Monday)
SiteERATO seminar room
Time15:30－17:30
SpeakerRyutaro Kurai
ProfileHatena
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateFebruary 24, 2011 (Thursday)
SiteERATO Seminar Room
Time16:00－18:00
SpeakerTsuyoshi Ueno
ProfileResearcher at ERATO Minato Discrete Structure Manipulation System Project
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateFebruary 7, 2011 (Monday)
SiteERATO Seminar Room
Time14:30－16:30
SpeakerDr. Paula Brito
ProfileAssociate Professor in Statistics and Data Analysis at the Faculty of Economics, The University of Porto
TitleIntroduction to Symbolic Data Analysis (An interaction movement between statistics and data processing)
AbstractIn classical data analysis, data is represented in a n×p matrix where n individuals (in rows) take exactly one value for each variable (in olumns). However, this model is too restrictive to represent complex data, which may comprehend variability and/or uncertainty. The need to consider data that contain information which cannot be epresented within the classical framewok lead to the development of Symbolic Data Analysis. Symbolic data extend the classical model, by allowing multiple, possibly weighted, values for each variable. New variable types have therefore been introduced, which allow us to represent ariability and/or uncertainty inherent to the data:multi-valued variables, nterval variables and modal variables. This approach is of particular and rowing interest in the analysis of huge sets of data, recorded in very large databases, when the units of interest are not the individual records (the microdata), but rather some second-level entities. For instance, in a database of credit card purchases, we are probably more terested in describing the behaviour of some person (or even some pre-defined class or group of persons) rather than each purchase by itself. By aggregating the purchase data for each person (or group), we obtain the information of interest; the observed variability for each person or within each group is preserved, and ith is of utmost importance. But are we still in the same framework when we allow for the variables to take multiple values? Are the definitions of basic statistical notions still so straightforward? What properties remain valid? In this talk we will discuss some issues that arise when trying to apply classical data analysis techniques to symbolic data. The central question of the evaluation of dispersion, and the consequences of different possible choices in the design of multivariate methods, will be addressed. Dispersion is a key issue in clustering, since the result of any lustering method depends heavily on the scales used for the variables; natural clustering structures can sometimes only be detected after an appropriate rescaling of ariables. The standardization problem has been addressed by De Carvalho, Brito & Bock, and three tandardization techniques for interval-type variables have been proposed. Furthermore, many exploratory ultivariate methodologies rely heavily on the notion of linear combination and on the properties of ispersion measures under linear transformations. This problem has been addressed in work of Duarte Silva & Brito in the context of linear discriminant analysis of interval data. Different approaches have been considered by various authors to address these and other questions and to propose a symbolic counterpart of statistical multivariate data analysis methods. As we can see, in recent years there has been a growing literature proposing methodologies for the analysis of symbolic data. However, most existing methods take a non-parametric descriptive approach. In a recent work, Brito and Duarte Silva focus on the analysis of interval data, for which probabilistic models are proposed and used. For modelling purposes, a parameterization consisting in representing each interval Iij=[lij , uij] by its midpoint cij=(lij+uij)/2 and range rij=uij-lij is adopted. The approach consists in modelling each pdimensional interval vector by a 2p-dimensional Normal or Skew-Normal distribution for the interval midpoints and log-ranges. One advantage of the Normal model is its analytical tractability and the possibility of straightforward applications of classical inference methods. On the other hand, the Skew-Normal model offers greater flexibility for the shape of the distributions. In the most general formulation we allow for non-zero correlations among all midpoints and log-ranges, but other particular cases of interest may also be taken into account. The proposed modelling may be applied to multivariate methodologies where a parametric distribution is to be assumed. Firstly, this odelling is employed in the context of (M)ANOVA techniques. This allows, in particular, assessing the relevance of different (interval) variables for a given partition on interval data. Secondly, a model-based clustering method (without fixing in advance the number of clusters), is eveloped, where a mixture distribution approach is followed, arameters and clusters are determined by maximum likelihood via the EM algorithm. This framework may be extended to other statistical methodologies, opening the way to inference approaches for symbolic data. In a most resent approach, quantile representation (Ichino, 2008) provides a common framework to represent symbolic data described by variables of different types. The principle is to express the observed variable values by some predefined quantiles of the underlying distribution. In the interval variable case, a distribution is assumed within each observed interval, e.g. uniform (Bertrand and Goupil, 2000) ; for a histogram-valued variable, quantiles of any histogram may be obtained by simply interpolation, assuming a uniform distribution in each class (bid); for categorical multi-valued variables, quantiles are determined from the ranking defined on the categories based on their frequencies. When quartiles are chosen, the epresentation for each variable is defined by the 5-tuple (Min, Q1, Q2, Q3, Max). This common representation then allows for a unified analysis of the data set, taking all variables simultaneously into ccount. In a numerical clustering context, the Ichino-Yaguchi dissimilarity is used to compare data units; hierarchical and pyramidal models, with several aggregation indices, may be applied and clusters are formed on the basis of quantile proximity. We also focus on a conceptual clustering approach. In this case, clusters are epresented, for each variable, by a mixture of the quantile-distributions of the merged clusters and then compared on the basis of the current quantile representation. The proposed hierarchical/pyramidal clustering model follows a bottom-up approach; at each step, the algorithm selects the two clusters with closest quantile representation to be merged. The newly formed cluster is then represented according to the same model, i.e., a quantile representation for the new cluster is determined from the uniform mixture cumulative distribution. Much remains however to be done. We have just rather briefly discussed some of the issues that arise when we leave the classical data framework and allow for more complex variable types. Usual properties, generally taken for granted, often do not apply any longer, and new concepts much be put forward. Among these, parametric statistical analysis is an important challenge. A whole world of problems still remains open, waiting to be explored.
Keywords : Symbolic data analysis, Interval data, Imprecise data,Standardization, Clustering, Discriminant Analysis, Parametric modelling of interval data, Statistical tests for interval data, Skew-Normal distribution, (M)ANOVA, Quantile representation
Connected siteTokyo, Kansai

Detail

DateJanuary 31, 2011 (Monday)
SiteERATO Seminar Room
Time13:15－15:00
SpeakerDr. Mathias Soeken
ProfileUniversity of Bremen
TitleFormal Verification of UML-based Specifications
AbstractIn the recent years, researchers started to enrich the classical
textbook specification of embedded systems by more formal descriptions
such as the Unified Modeling Language (UML). This enables to lift the
starting point for verifcation techniques to the specification level
and leads to an improved design flow, where crucial flaws can be
detected, even if no executable implementation is available. However,
research on efficient solving techniques for the respective
verification tasks is just at the beginning.
In the talk, the improved design flow based on specifications enriched
with formal descriptions is introduced. We illustrate how early design
flaws become evident within this flow and briefly review algorithms
aimed at detecting them. Furthermore, first experimental results are
presented. Overall, the aim of the talk is to provide an introduction
into this emerging area and to present first results.
Connected siteTokyo

Detail

DateJanuary 28, 2011 (Friday)
SiteERATO Seminar Room
Time16:00－17:30
SpeakerHiroaki Iwashita
ProfileERATO Minato Discrete Structure Manipulation System Project
Researcher
TitleImproving Simulation Coverage of Metastability Effects in Clock Domain　Crossing Verification

Abstract

Detail

DateJanuary 21, 2011 (Friday)
SiteTokyo Satellite Lab
Time15:30－17:30
SpeakerMasaaki　Kotera
ProfileKyoto University
Title
Abstract
Connected siteSapporo, Kansaii

Detail

DateJanuary 11, 2011 (Tuesday)
SiteERATO Seminar Room
Time15:30－17:00
SpeakerMakoto Kanazawa
ProfileNational Institute of Informatics
TitleA Chomsky-Sch\"{u}tzenberger-Weir Representation Theorem for Simple Context-Free Tree Grammars
AbstractAnalogues of the Chomsky-Sch\"{u}tzenberger Theorem for context-free
languages have been obtained for two distinct classes of languages,
both of which have sometimes been identified with the informal notion
of "mildly context-sensitive" languages.

In 1988, Weir showed that each tree-adjoining language is a
homomorphic image of the intersection of two Dyck languages over 2n
pairs of brackets and a regular set, where the two Dyck languages are
related by a permutation of the alphabet that maps

[_{2i+1} ]_{2i+1} [_{2i+2} ]_{2i+2}

to

[_{2i+1} [_{2i+2} ]_{2i+2} ]_{2i+1}

for each i = 0,...,n-1.

Recently, Yoshinaka et al. (2010) obtained a representation theorem
for each level in the two-dimensional hierarchy of multiple
context-free languages, which subsumes tree adjoining languages and
much more. They introduced the notion of "multiple Dyck languages"
(for each dimension and rank), and showed that every multiple
context-free language of a given dimension and rank is a homomorphic
image of the intersection of a multiple Dyck language and a regular
set.

In this talk, I generalize Weir's representation theorem for the
tree-adjoining grammars to the simple (i.e., non-deleting and
non-duplicating) context-free tree grammars, which correspond at the
level of string languages to the "well-nested" multiple context-free
grammars. The well-nested multiple context-free languages form a
large subclass of the multiple context-free languages with some
interesting distinguishing properties. In this representation, each
string language generated by a context-free tree grammar of rank m is
obtained as a homomorphic image of the intersection of two Dyck
languages over (m+1)n pairs of brackets and a regular set, where the
two Dyck languages are related by a permutation of the alphabet that
maps

[_{mi+1} ]_{mi+1} [_{mi+2} ]_{mi+2} ... [_{mi+m} ]_{mi+m}

to

[_{mi+1} [_{mi+2} ]_{mi+2} [_{mi+3} .... ]_{mi+m} ]_{mi+1}

for each i = 0,...,n-1. This is obtained as a consequence of a
representation theorem at the level of tree languages of simple
context-free tree grammars, where a tree analogue of Dyck languages
plays the role of the usual Dyck languages.
Connected siteTokyo, Kansai

Detail

DateJanuary 7, 2011 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerKimihito Ito
Profile
TitlePrediction of amino acid substitutions on the hemagglutinin　molecules of influenza A viruses

AbstractHuman influenza viruses mutate from time to time, causing annual
epidemics worldwide. Given the high mutation rate of the viral gene, it
is difficult to select an effective vaccine strain prior to each influenza season. In order to elucidate the pattern of viral evolution, we Prediction of amino acid substitutions on the hemagglutinin molecules of influenza A viruses introduce a bioinformatics technology that analyzes a large number of epidemic strains in a multidimensional space. We found that the relative sequence distances among past epidemic strains could be predicted by a mathematical model. Retrospective tests for 12 years showed that the model could predict the direction of viral evolution with high accuracy.
Through these technologies, we investigate the past, current and future evolution of influenza A viruses.

Detail

DateDecember 17, 2010 (Friday)
SiteKansai Satellite Lab.
Time15:30－00:01
SpeakerGuan-Cheng Li
ProfilePhD student in Computer Science, University of California, Berkeley
TitleMining Psychology from English News and Applications on Finance.

AbstractOur perceptions of the world are largely shaped by news media.
Understanding the psychology of how media portray certain words is a
critical step towards assessing media's influence on those
perceptions. We implement a system which analyzes the 'image' of a
given query word in a given corpus of news texts by producing a list
of other words with which this query is strongly associated, which
will be used to construct a covariance matrix amongst financial
assets, based on their co-appearances on news. The presentation will
cover theories of computational linguistics, machine learning,
sentiment analysis, and implementation issues in real time news
analysis.
Connected siteSapporo, Tokyo

Detail

DateNovember 19, 2010 (Friday)
Site
Time00:01－00:01
SpeakerProf.　Kai Ming Ting
ProfileGippsland School of Information Technology, Monash University　 Associate Professor
TitleMulti-Dimensional Mass Estimation and Mass-based Clustering

Abstract
Connected siteSapporo, Tokyo

Detail

DateNovember 15, 2010 (Monday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerProf. Adnan Darwiche
Profile
TitleEfficient Representations of Boolean Functions: The View from Knowledge Compilation
AbstractWe provide an overview of the area of knowledge compilation, which is
concerned with the systematic study of efficient representations of
propositional knowledge bases (aka, Boolean functions). We discuss the
two key dimensions of knowledge compilation: succinctness of the
representation, and the set of operations it supports in polytime. We
focus on two key representations: The well-studied Ordered Binary
Decision Diagram (OBDD) and the more recent Decomposable Negation
Normal Form (DNNF). We highlight the relationship between these two
representations, while stressing some of the more recent developments
on DNNFs. This includes lower and upper bounds on the size of DNNFs
and how they relate to similar bounds on OBDDs; the notion of a vtree
and how it generalizes the notion of a variable order; and the newly
discovered notion of "interaction function" and the role it plays in
the decomposition and efficient representation of Boolean functions.
Connected siteTokyo, Kansai

Detail

DateNovember 12, 2010 (Friday)
SiteKansai Satellite Lab.
Time15:30－17:30
SpeakerJoe Suzuki
ProfileOsaka University
Title
Abstract
Connected siteSapporo, Tokyo

Detail

DateNovember 5, 2010 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerRyo Yoshinaka
ProfileResearcher, JST ERATO Minato Discrete Structure Manipulation System Project
TitleIntroduction to Sequence BDD and Comparison with Previous Methods
Abstract
Connected siteTokyo, Kansai
Time15:30－17:30
SpeakerShuhei Denzumi
ProfileHokkaido University
TitleSeqBDD : Introduction to Sequence BDD
Abstract
Connected siteTokyo, Kansai

Detail

DateOctober 29, 2010 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerTakeru Inoue
Profile
Title
Abstract

Detail

DateOctober 22, 2010 (Friday)
SiteERATO Seminar Room
Time15:30－17:01
SpeakerShin-ichi　Minato
ProfileProfessor at Hokkaido University and Research Director of JST ERATO Minato Discrete Structure Manipulation System Project
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateSeptember 27, 2010 (Monday)
SiteERATO Seminar Room
Time10:00－12:30
SpeakerShohei Hido
ProfileIBM Research - Tokyo
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateSeptember 27, 2010 (Monday)
Site
Time10:00－12:30
SpeakerTakashi Imamichi
ProfileIBM Research - Tokyo
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateSeptember 27, 2010 (Monday)
SiteERATO Seminar Room
Time10:00－12:30
SpeakerTsuyoshi　Ide
ProfileIBM Research - Tokyo
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateSeptember 16, 2010 (Thursday)
SiteERATO Seminar Room
Time15:00－17:00
SpeakerTakashi Horiyama
Profile
Title
Abstract

Detail

DateSeptember 2, 2010 (Thursday)
SiteERATO Seminar Room
Time15:30－17:00
SpeakerProf. Randy Goebel
Profile
TitleChallenges for a theory of visualization: what is semantic symmetry?
AbstractWhile there is no theory of visualization, there should be. Such a　theory would provide a framework to assess a variety of information　visualization techniques, to understand their comparative value in　helping humans to draw inferences on large data. One simple concept
related to a theory of visualization is semantic symmetry, which can be
considered as the property that change in one representation space (e.g.,　a visual vocabulary space) can be accurately propogated to another space　(e.g., a numeric tabular space). We explain the idea of semantic　symmetry, and its potential role in a theory of visualization.
Connected siteKansai

Detail

DateAugust 26, 2010 (Thursday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerYasuo Tabei
ProfileResearcher, JST ERATO Minato Discrete Structure Manipulation System Project
Title
Abstract
Connected siteKansai

Detail

DateAugust 23, 2010 (Monday)
SiteERATO Seminar Room
Time14:30－16:00
SpeakerTsuyoshi　Shibata
Profile
Title
Abstract

Detail

DateAugust 20, 2010 (Friday)
SiteERATO Seminar Room
Time10:30－12:00
SpeakerShigeru Yamashita
ProfileProfessor, Ritsumeikan University
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateAugust 20, 2010 (Friday)
SiteERATO Seminar Room
Time10:30－12:00
SpeakerByung-Soo Choi
Profile
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateAugust 6, 2010 (Friday)
SiteERATO Seminar Room
Time10:00－12:00
SpeakerTakeaki Uno
ProfileNational Institute of Informatics
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateAugust 6, 2010 (Friday)
SiteERATO Seminar Room
Time13:00－16:00
SpeakerAkihiro Yamamoto
Profile
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateAugust 4, 2010 (Wednesday)
SiteERATO Seminar Room
Time10:00－14:00
SpeakerRyuhei Uehara
ProfileJapan Advanced Institute of Science and Technology
Title
Abstract

Detail

DateAugust 3, 2010 (Tuesday)
SiteKansai Satellite Lab.
Time15:30－17:30
SpeakerShigeru Yamashita
ProfileRitsumeikan University
TitleIntroduction to Grover Algorithm
Abstract
Connected siteSapporo, Tokyo

Detail

DateAugust 3, 2010 (Tuesday)
SiteKansai Satellite Lab.
Time15:30－17:30
SpeakerProf. Byung-Soo Choi
ProfileEwha Womans University
TitleGrover Search and its Applications
Abstract
Connected siteSapporo, Tokyo

Detail

DateJuly 23, 2010 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerTakuya Kida
Profile
Title
Abstract

Detail

DateJuly 9, 2010 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerYasuyuki Shirai
ProfileERATO Minato Discrete Structure Manipulation System Project
Research Manager/Researcher
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateJuly 2, 2010 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerToshiki Saitoh
ProfileERATO Minato Discrete Structure Manipulation System Project
Researcher
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateJune 18, 2010 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerJun Kawahara
ProfileERATO Minato Discrete Structure Manipulation System Project
Researcher
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateJune 4, 2010 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerRyo Yoshinaka
ProfileERATO Minato Discrete Structure Manipulation System Project
Researcher
TitleChomsky-Schützenberger-Type Characterization of Multiple Context-Free Languages
AbstractIt is a well-known theorem by Chomsky and Sch◯tzenberger (1963) that every context-free language can be represented as a homomorphic image of the intersection of a Dyck language and a regular language.
This paper gives a Chomsky-Sch◯tzenberger-type characterization for multiple context-free languages, which are a natural extension of context-free languages, with introducing the notion of multiple Dyck languages, which also are a generalization of Dyck languages.
Connected siteTokyo, Saitama University

Detail

DateMay 21, 2010 (Friday)
SiteERATO Seminar Room
Time15:30－17:30
SpeakerShin-ichi　Minato
ProfileProfessor at Hokkaido University and Research Director of JST ERATO Minato Discrete Structure Manipulation System Project
Title
Abstract
Connected siteTokyo, Kansai

Detail

DateMay 14, 2010 (Friday)
Site
Time15:30－18:00
SpeakerZin Arai
ProfileAssistant Professor, Creative Research Institution "Sousei", Hokkaido University
Title
Abstract