Showing 1–2 of 2 results for author: Ohira, T

A Neural Network model with Bidirectional Whitening

Authors: Yuki Fujimoto, Toru Ohira

Abstract: We present here a new model and algorithm which performs an efficient Natural gradient descent for Multilayer Perceptrons. Natural gradient descent was originally proposed from a point of view of information geometry, and it performs the steepest descent updates on manifolds in a Riemannian space. In particular, we extend an approach taken by the "Whitened neural networks" model. We make the white… ▽ More We present here a new model and algorithm which performs an efficient Natural gradient descent for Multilayer Perceptrons. Natural gradient descent was originally proposed from a point of view of information geometry, and it performs the steepest descent updates on manifolds in a Riemannian space. In particular, we extend an approach taken by the "Whitened neural networks" model. We make the whitening process not only in feed-forward direction as in the original model, but also in the back-propagation phase. Its efficacy is shown by an application of this "Bidirectional whitened neural networks" model to a handwritten character recognition data (MNIST data). △ Less

Submitted 24 April, 2017; originally announced April 2017.

Comments: 16pages

Journal ref: In: Rutkowski L., Scherer R., Korytkowski M., Pedrycz W., Tadeusiewicz R., Zurada J. (eds) Artificial Intelligence and Soft Computing. ICAISC 2018. Lecture Notes in Computer Science, vol 10841. Springer, Cham

Chases and Escapes, and Optimization Problems

Authors: Toru Ohira

Abstract: We propose a new approach for solving combinatorial optimization problem by utilizing the mechanism of chases and escapes, which has a long history in mathematics. In addition to the well-used steepest descent and neighboring search, we perform a chase and escape game on the "landscape" of the cost function. We have created a concrete algorithm for the Traveling Salesman Problem. Our preliminary t… ▽ More We propose a new approach for solving combinatorial optimization problem by utilizing the mechanism of chases and escapes, which has a long history in mathematics. In addition to the well-used steepest descent and neighboring search, we perform a chase and escape game on the "landscape" of the cost function. We have created a concrete algorithm for the Traveling Salesman Problem. Our preliminary test indicates a possibility that this new fusion of chases and escapes problem into combinatorial optimization search is fruitful. △ Less

Submitted 1 December, 2014; originally announced December 2014.

Comments: 3 pages, 4 figures. To appear in the Proceedings of the International Symposium on Artificial Life and Robotics (AROB20th), Beppu, Oita Japan, January 21-23, 2015

Journal ref: Artificial Life Robotics (2015) 20: 257