Machine Learning for Science

Our aim is to develop new machine learning methods and algorithms for analyzing Big Data in Science. The techniques are applied to find new scientific insights in the collaborations with application domains.



Postdoc positions are available

Postdoc positions to develop machine learning methods for multi-omics data analysis and its application to real biology and medical data are available. Please contact to

Posted Nov. 22, 2016

Starting a collaboration with National Cancer Center and Preferred Networks

Our proposal to develop an AI system for cancer treatments has been accepted in JST-CREST project (Leader: Prof. Ryuji Hamamoto), and we start the collaboration of the development of medical AIs. The press conference was held.

Posted Nov. 29, 2016

IIBMP 2016

We organized Japanese Bioinformatics community meeting, IIBMP 2016. More than 500 researchers were attended, and more than 30 companies were supported. Thank you very much!

Posted Oct 1, 2016

Starting plant science projects

Our proposals for the analysis of hybrid species and its application to crop sciences have been accepted to KAKENHI (Leader: Jun Sese) and JST-CREST (Leader: Prof. Kentaro K. Shimizu from Kanagawa City Univ and Zurich Univ) in this year. We'll perform in natura multi-omics analysis (genome, expression, methylation and images) over multiple sites.

Posted Sep. 16, 2016

New Team!

We formed a new team named Machine Learning Research Group in Artificial Intelligence Research Center, AIST. In the team, we will develop new machine learning methods as well as their applications to bio/medical informatics and material sciences.

Posted April 1, 2016

Structural Variation Detection

Our paper on structural variation detection is published from Nucleic Acids Research. Its Japanese press release is out from AIST. This is collaborative work with Prof. Takeda's group at Osaka University Medical School. Software is available

Posted Feb. 2, 2016

Bioinformatics Book

A Japanese book entitled "Machine learning in bioinformatics - multiple testing and estimator design" was published from Kodansha scientific. Link to Thank you for collaborators!

Posted Dec. 9, 2015

Multiple testing with GPGPU for GWAS

Efficient parallelizing method for permutation-based multiple testing procedure with GPGPU is accepted in ACM BCB-2015. This method is specialized on the case when millions of tests are required such as GWAS data.

Posted Sep. 11, 2015


Jun will make a presentation in IPSJ-ONE, which consists of twenty TED-style lightning talks presented by top-level young computer science researchers in Japan. It is to be broadcast by Nico-Nico Douga. Check it out!

Posted Mar. 1, 2015

Tokyo Workshop on Statistically Sound Data Mining

Prof. Tsuda (U. Tokyo) and Jun organize an international workshop about statistically sound data mining. This workshop aims to bring together prominent researches between data mining and life science to find new direction in newly rising field of statistically sound data mining.

Posted Feb. 1, 2015

Dr. Jun Sese received Oxford Journal-JSBi Prize

Dr. Jun Sese received Oxford Journal-JSBi Prize, which is the highest impact young investigator award in Japanese Society for Bioinformatics. Thank you very much for your continuous support.

Posted Oct. 4, 2014

We got three research awards in IIBMP 2014

In the annual conference of Informatics in Biology, Medicine and Pharmacology (IIBMP) 2014, we got three research awards. Dr. Koichi Yamagata received IIBMP 2014 excellent research award for his genomic stractural variation detection research, and Mr Hanyoung Kim and Yuki Saito, who are master course students of Tokyo Tech, are received JSBi excellent research awards for their algorithmical and statistical researches, respectively. Congrats!

Posted Oct. 4, 2014

Expression analysis of hybrid species with NGS

Joint work with Shimizu Lab. has been published in Nucleic Acids Research. This paper introduced a novel method for comprehensive expression analysis of hybrid species with NGS, including a new statistical method. Also this paper provided two genomes of Arabidopsis lyrata and Arabidopsis halleri.
All scripts are available.

Posted Jan. 14, 2014

Causality inference from E. coli expression data

Joint work with Dr. Makoto Yamada and Prof. Masashi Sugiyama has been accepted in Machine Learning. In the paper, we predicted regulation direction of genes in E. coli from expression data, not only developed a novel causal learning method.

Posted Nov. 14, 2013

Combinatorial feature discovery with random permutations

Our paper on statistically significant feature discovery with random permutations has been accepted in IEEE BIBM 2013 as a regular paper. We proposed an efficient calculation of Westfall-Young multiple testing correction procedure, especially to discover statistically significant combinations of features. Acceptance ratio is 19.8%.

Posted Nov. 13, 2013

Special Article on Nature Japan

Nature Japan published a special article on our multiple testing correction project (Japanese).

Posted Aug. 23, 2013

Statistics to Discover Salient Combinations

We published a paper on PNAS (Software)that proposed a multiple testing correction method to discover epistatic combinations with statistical significance, named LAMP. We issued the press release (Japanese).

Posted July 24, 2013

© 2016 Machine Learning Team. Back to Top