Title :
Worst case analysis of weight inaccuracy effects in multilayer perceptrons
Author :
Anguita, Davide ; Ridella, Sandro ; Rovetta, Stefano
Author_Institution :
Dept. of Biophys. & Electron. Eng., Genoa Univ., Italy
fDate :
3/1/1999 12:00:00 AM
Abstract :
We derive a method for the analysis of weight quantization effects in multilayer perceptrons based on the application of interval arithmetic. Differently from previous results, we find worst case bounds on the errors due to weight quantization, that are valid for every distribution of the input or weight values. Given a trained network, our method allows us to easily compute the minimum number of bits needed to encode its weights
Keywords :
multilayer perceptrons; pattern classification; quantisation (signal); interval arithmetic; trained network; weight inaccuracy effects; weight quantization effects; worst case analysis; Algorithm design and analysis; Arithmetic; Computer aided software engineering; Computer networks; Multilayer perceptrons; Noise robustness; Nonhomogeneous media; Performance analysis; Quantization; Registers;
Journal_Title :
Neural Networks, IEEE Transactions on
