Adaptive systems can adapt and learn characteristics and
functions, which are required in the unknown or time varying environments.
Among these systems, especially adaptive filters and neural networks are
our main research subjects.
Adaptive filters are used for estimating the signal of
interest under nonlinear distortion, noisy and interference environments.
Concrete subjects are fast and stable learning algorithms, noise chancellors
and echo chancellors, blind channel identification and equalization, nonlinear
adaptive filters.
Neural networks are effectively applied to prediction
and diagnosis of many kinds of phenomena and systems. The following concrete
subjects are studied: network structure learning, generalization with sparse
training data, applications to prediction, diagnosis and pattern classification
and blind source separation.
These researches will be connected to advance in information
quality and development of intelligent and flexible systems.
適応･学習能力を有し、ダイナミカルに変化するシステム
･適応フィルタ／ニューラルネットワーク
･システムの最適構成
･学習アルゴリズム
･ダイナミクス解析
･通信･信号処理･知識処理等への応用
･雑音に埋もれた信号の抽出
･歪んだ信号の復元
･混成信号からの分離推定