WIT Press

The Performance Of Various Learning Rates For An Unsupervised Neural Network

Price

Free (open access)

Paper DOI

10.2495/AI970321

Volume

19

Pages

20

Published

1997

Size

414 kb

Author(s)

C.K. Lee & C.H. Chung

Abstract

In this paper, we shall investigate the effect of learning rate and threshold to an unsupervised neural network when applied to an inspection process. The network we use is the Learning by Experience (LBE)[1]. Here, we analyse the effect based on a performance index. Experimental results are included when this neural network is applied to IC leadframe inspection. 1. Introduction Whenever we want to apply an unsupervised neural network for inspection, we first have to set up the initial values of some parameters before we can progress. In our case, these parameters are the learning rate and threshold. For the learning rate, its purpose is to adapt the weight vector to a new pattern. The threshold means the acceptance criterion for a certain part. This paper will discuss the

Keywords