Geometric Properties of Quasi-Additive Learning Algorithms*This study is supported in part by Grant-in-Aid for Scientific Research (14084210, 15700130, 18300078) from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

  • Authors:

  • Kazushi Ikeda

  • Affiliations:

  • The author is with Kyoto University, Kyoto-shi, 606-8501 Japan. E-Mail: kazushi@i.kyoto-u.ac.jp

  • Venue:
  • IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

The family of Quasi-Additive (QA) algorithms is a natural generalization of the perceptron learning, which is a kind of on-line learning having two parameter vectors: One is an accumulation of input vectors and the other is a weight vector for prediction associated with the former by a nonlinear function. We show that the vectors have a dually-flat structure from the information-geometric point of view, and this representation makes it easier to discuss the convergence properties.