Vector Generations in Neural Network Computations

Naohiro Ishii, Toshinori Deguchi, Masashi Kawaguchi, and Hiroshi Sasaki

Keywords

neural network computation, parallel computation in neural net, vector generation

Abstract

Data pathways are important in layered neural networks. The problem is how to classify information pathways in the network computations. First, the architecture of the biological asymmetric network with odd-even (or evenodd) order nonlinearities is analyzed for the network computations. The stimulus with a mixture distribution is useful to evaluate their network processing ability for the movement direction and its velocity, which generate a vector. Then, white noise analysis is applied to solve the problem. Thus, the characterized equation is derived in the network computations. which evaluates the processing ability of the network. Second, the movement velocity is derived, which is represented in Wiener kernels of the network computations. Thus, the information pathways for characterizing the ability of the movement detection are classified for the layered neural networks computations.

Important Links:



Go Back