MURATA, Noboru

Recent progress of computers and storage systems allows us to deal with massive data quite efficiently. Huge datasets are collected with a great variety of sensors and machine learning techniques are expected to successfully retrieve important information from those datasets.  Our research interests include the theoretical aspects of machine learning focusing on the dynamics and statistical properties of learning algorithms such as neural networks, boosting and bagging, sparse representation and information geometry.  Also we are interested in applying our theories to data analysis such as prediction of energy demands and anomaly detection of industrial plants.