Graduate School of Science and Engineering Information and Computer Science

Co-Creation Informatics Laboratory

Research on pattern recognition, machine learning, knowledge discovery, signal processing, and human-computer collaboration

Staff

OHSAKI Miho
[Professor]

Acceptable course
Master's degree course	✓
Doctoral degree course	✓

Telephone : +81-774-65-6468
mohsaki@mail.doshisha.ac.jp
Office : KE-111
Prof. Ohsaki’s publications

SHIRAHAMA Kimiaki
[Associate Professor]

Acceptable course
Master's degree course
Doctoral degree course

Telephone : +81-774-65-7567
kshiraha@mail.doshisha.ac.jp
Office : KE-110
Prof. Shirahama’s publications

In our laboratory, we develop technologies for machine learning, knowledge discovery, and multimedia understanding with the aim to improve human collaboration and productivity by supporting intellectual activities. We also apply the findings of the fundamental research to the fields of medical informatics and education. As a research environment, high performance cloud service platforms such as AWS and computation servers equipped with many cores and GPUs are available to efficiently carry out computational experiments and simulations. The present main research themes are as follows.

*******************************************
(1) Development of Machine Learning and Knowledge Discovery Methods
*******************************************
[ Themes on Time Series ]
(1)-1: Methods to analyze biomedical signals based on signal processing and time series deep learning.
(1)-2: Methods to generate basis functions based on deep unfolding and self-supervised learning.

[ Themes on imbalance ]
(1)-3: Deep neural networks for imbalanced data classification based on the confusion matrix.
(1)-4: Satellite image recognition using the imbalanced data classifier.

[ Themes on dependence and causality ]
(1)-5: Methods for nonlinear dependence discovery based on neural networks with the L1 regularization.
(1)-6: Methods for nonlinear causality discovery by the expansion of dependence discovery methods.

*******************************************
(2) Development of Multimedia Understanding Methods
*******************************************
[ Themes on Cognitive System Modeling ]
(2)-1: Matching of image regions and words/phrases based on attention mechanism.
(2)-2: Modeling of memory mechanism during video viewing.
(2)-3: Data augmentation by modeling human's imagination process.

[ Themes on Multimodal Analysis ]
(2)-4: Integration, imputation, and correspondence among data of different modalities.
(2)-5: Building a question-answering system based on data from multimodalities.
(2)-6: Intention inference through natural interaction between a user and a system.

*******************************************
(3) Development and Application of Intellectual Activity Support Systems to Medical Informatics and Education
*******************************************
[ Themes on Medical and Health Care ]
(3)-1: Brain occlusion inference using pulse waves.
(3)-2: Bone quality inference using ultrasonic waves.
(3)-3: Biomedical signal analysis with medical explainability.
 
[ Themes on Education ]
(3)-4: Data mining for STEM education.
(3)-5: Verification of knowledge consistency among different information sources.