What question did this study set out to answer?

This research aims to develop a penalty function for effectively pruning feedforward neural networks by eliminating unnecessary weights.

A Penalty-Function Approach for Pruning Feedforward Neural Networks

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial evaluates weight elimination in neural networks, suggesting effective pruning techniques.

This research aims to develop a penalty function for effectively pruning feedforward neural networks by eliminating unnecessary weights.
Proposes a penalty function with two terms: one to discourage unnecessary connections and another to limit excessively large weight values.
Provides criteria for eliminating weights based on the penalty function.
Tests the effectiveness of the pruning technique on the contiguity, parity, and monks problems.
Demonstrates that pruned networks have fewer connections than previously reported for these problems.
Achieves significant weight reduction while maintaining performance across the tested problems.

Rudy Setiono (1997) studied this question.

View Full Paper

Ask AI

Bookmark