Research Themes

Laboratory of Large-Scale Knowledge Processing

Themes

High Performance Algorithms and Applications to Artificial Intelligence and Knowledge Processing

An algorithm is the sequence of computational instructions written in a computer program.  In many cases, we can greatly improve computational performance simply by improving the underlying algorithm.  Many programs use basic mathematical structures such as sets, logic, proofs, graphs, permutations, combinations and probabilities.  We call these discrete structures.  Algorithms for discrete structures are a fundamental technique with many applications and tremendous impact, necessary in many aspects of modern society including hardware and software design, fault analysis in massive systems, constraint satisfaction problems, data mining and knowledge discovery, machine learning and automatic classification, bioinformatics and web analysis.

ZDD

Knuth1   Knuth2

Professor Minato introduced and named ZDDs (Zero-Suppressed Binary Decision Diagrams), an algorithmic technique used with discrete structures.  ZDDs compress and index data on discrete structures without modifying the underlying meaning, and provide many efficient operations.  ZDDs are now widely-used, and ZDD-based techniques are introduced in-depth in the so-called Bible of algorithms — The Art of Computer Programming (D. E. Knuth, volume 4-1).  This is a first for a Japanese researcher, demonstrating the impact ZDDs have had.

Mirai1

Mirai2       image10

This research on ZDDs was introduced to schoolchildren and the general public in an exhibit at Miraikan (National Museum of Emerging Science and Innovation), teaching combinatorial explosion and the importance of algorithms.  As part of the exhibit, Prof. Minato supervised the creation of an animation that has been viewed over 1.8 million times on YouTube.  It has been used in high school and college classes, and the number of views continues to rise.

Network          Map

Our group working on high performance algorithms for processing discrete structures uses state-of-the-art algorithmic theory and the latest techniques from machine learning, artificial intelligence, data mining, statistics, system optimization, and information security.  We seek to have an impact on a wide range of real-world topics, including applications in designing and analyzing eletrical power, communication, transportation, business transactions, disaster prevention, and also big data analysis related to materials and life sciences.

Faculty

Shin-ichi Minato
Shin-ichi Minato
Position
Professor
Specialized field
Binary decision diagrams (BDD), logic functions, logic circuits, zero-suppressed binary decision diagrams (ZDD), sets of combinations, discrete structure manipulation, knowledge discovery and data mining, enumeration algorithms, succinct data structures, graph theory, probabilistic reasoning and knowledge representation
Hobby
Orchestral and chamber music (violin performance), sports, making electronic devices

Become a person who will propose original and useful ideas. Make your effort to consider the insufficient points in the current technology and how to improve them.

Ichigaku Takigawa
Ichigaku Takigawa
Position
Associate Professor
Specialized field
Machine learning and its applications, machine learning for discrete structures (especially graphs, networks, trees), enumeration and search of substructural patterns, knowledge discovery from life science and materials science data, data analysis for molecular biology and drug development, bioinformatics
Hobby
Architecture, music, snow removal for curbside parking, ontological and epistemological thinking

Think hard, work hard and enjoy it!