A Simple Randomized Quantization Algorithm for Neural Network Pattern Classifiers
暂无分享,去创建一个
This paper explores some algorithms for automatic quantization of real-valued datasets using thermometer codes for pattern classiication applications. Experimental results indicate that a relatively simple randomized thermometer code generation technique can result in quantized datasets that when used to train simple perceptrons, can yield generalization on test data that is substantially better than that obtained with their unquantized counterparts.
[1] Rajesh Parekh,et al. Analysis of Decision Boundaries Generated by Constructive Neural Network Learning Algorithms , 1995 .
[2] Vasant Honavar,et al. Generative learning structures and processes for generalized connectionist networks , 1993, Inf. Sci..
[3] Stephen I. Gallant,et al. Perceptron-based learning algorithms , 1990, IEEE Trans. Neural Networks.