[Japanese version]
NTT Communication Science Laboratories
Innovacative Communication Labotratory
Learning and Intelligent Systems Research Group


Futoshi Naya (Group Leader)
Kazuo Aoyama
Akinori Fujino
Yoshinari Shirai
Shin Mizutani
Jun Mumramatsu
Yasue Kishino
Hitoshi Shimizu
Takashi Hattori
Takuma Otsuka
Yoichi Chikahara


Naonori Ueda (NTT Fellow, Ueda Research Laboratory)
Takeshi Yamada (Research Planning Section)
Hiroshi Sawada (Innovacative Communication Labotratory)
Tomoharu Iwata (Ueda Research Laboratory)
Katsuhiko Ishiguro (Mirai Translate, Inc.)
Masakazu Ishihata (Hokkaido University)
Koh Takeuchi (Ueda Research Laboratory)
Mathieu Blondel (Ueda Research Laboratory)
Takayuki Suyama (ATR)
Hitoshi 'keen' Sakano (NTT DATA)
Kazuhiko Shinozawa (Osaka Kyoiku University)
Akisato Kimura
Makoto Yamada (Research Associate)
Kazumi Saito (University of Shizuoka)
Kenichi Arai (Signal Processing Research Group)
Tatsushi Matsubayashi (NTT EV Labs.)
Shuhei Kuwata (NTT DOCOMO)
Noriaki Kawamae (NTT COMWARE)
Akinori Abe (Chiba University)
Daichi Mochihashi (The Institute of Statistical Mathematics)
Albert (Ching-man) Au Yeung (Hong Kong Applied Science and Technology Research Institute)
Takuya Maekawa (Osaka University)
Koichi Fujiwara (Kyoto University)
Yasuko Matsubara (Kumamoto University)
Yasushi Sakurai
Makoto Yamada
Yutaka Yanagisawa
Machiko Toyoda (NTT Software Innovation Center)

  Research Overview

The Learning and Intelligent Systems Research Group is committed to the research of analyzing complex phenomena emergent in the real and cyber worlds by using statistical machine learning, data mining, data stream analysis and sensor networks.

Discover and visualize latent structures in relational data

In our modern society, there is a huge amount of information available, and people have to deal with more information than they are able to process to make sensible decisions. We are pursuing statistical machine learning approaches to establish highly sophisticated data-mining technologies to analyze, organize, classify and visualize information. By understanding generative and evolution processes of the information, we can discover important but latent causes for observable interactions and relations in the data, and make predictions. In short, our aim is to make full use of information as knowledge. In particular, our research topics include "visualization of large-scale relational data" that enables us to visualize latent structures and characteristics of relational data, "fast similarity search for large-scale data" by organizing data as a network, "modeling purchase behavior" and "multiple object tracking and clustering" both of which extract and model aspects of human behaviors.

Future Directions

All the social phenomena occur as results of accumulated interactions between humans or human and information in various levels. We believe that we can unravel key mechanism of the generative processes of those social phenomena through the analysis and understanding of those interactions. For example, by analyzing the information flows and interactions on the whole Internet, we may able to predict future growth, anticipate the trend and early sign of anomaly, and visualize whole Internet structure. This is like a weather forecast system for the Internet. Furthermore, by analyzing all possible sensory data that monitor every aspect of human activities, we may able to categorize human daily and social activities into patterns and predict future actions that enable us to avoid potential risks, find others who have similar activity patterns, be advised next possible actions and improve technical skills or expertise. In short, we will be able to build a total navigation system of human daily life.

Visualizing large-scale relational data

We study visualizing large-scale networks with hundreds of thousands of nodes fast and accurately, based on a force-directed model and inspired by a hierarchical approach known in the astronomy.

We also study simultaneous extracting and visualizing both documents and semantic "topics" of the documents. PLSV

Fast similarity search for large-scale data

We study efficient similarity search methods for large-scale data that utilizes a network with a small-world property as an indexing structure.
fast similarity search

Extracting patterns from human behaviors

We try to model human behavior of purchasing items to discover innovative users who purchase items earlier than others and to improve customer lifetime value by recommending items that will help extending subscription period. We also study simultaneous tracking objects such as humans and extracting the number of objects, positions and dynamics patterns in a movie at the same time.
Positions available: We recruit talented researchers who are interested in Statistical Machine Learning and Data Mining. We have positions for new graduates and mid-careers as well as contract positions (RA = Post-doc and RS). For more details, please visit Career Opportunities page.

Last modified on: Nov 30, 2009