Matsuyama Laboratory image

Yasuo MATSUYAMA
Professor Emeritus and
Honorary Researcher of
Research Institute of
Science and Engineering,
(Computer Science and Communications Engineering Major)

Yasuo MATSUYAMA

Profile (For details, click the link of "Professor")

Professor Matsuyama is best known as the founder of the α-EM algorithm (alpha Expectation Maximization algorithm). He has contributed to the machine learning and computational intelligence with novel algorithms and applications. This field brings about the integration of symbols and patterns for IoCT (Internet of Collaborative Things).

His contributions to the above started from two doctoral studied in his early days. The first is the studies on stochastic modeling of neural spike trains which comprises the pulse-frequency-modulation (Doctor of Engineering from Waseda University). Another one is on the studies on the process distortion measures and signal processing that led to the very low-rate speech compression by clustering algorithms (Ph.D. from Stanford University).

Supported by the experience in these two fields, Professor Matsuyama presented algorithms of machine learning and computational intelligence that interacts living bodies and machines.

Contributions by Professor Matsuyama are grouped as follows.

  • Theory of generalized informational divergences for the likelihood optimization
  • Alpha-EM algorithm (alpha Expectation Maximization algorithm)
  • Alpha-HMM algorithm (alpha Hidden Markov Model)
  • Rapid ICA algorithm (Rapid Independent Component Analysis algorithm)

  • IoCT (Internet of Collaborative Things)
  • Collaboration of machine learning and blockchain

  • Harmonic competition and self-organizing optimization
  • Multiple descent cost copmetitive learning algorithm and injection of external intelligence
  • Speech compression by vector quantization (very low-rate data compression by clustering)

  • Similar image retrieval system (RIM, Retrival-aware IMage format)
  • Similar video retrieval system and the M-distance

  • Humanoid operation by brain signals
  • Authentification by brain signals

  • Neural networks and spiking neuron modeling
  • Bioinofrmatics (promoter recognition)