Compete to Compute
Rupesh Kumar Srivastava, Jonathan Masci, Sohrob Kazerounian, Faustino J. Gomez, Jürgen Schmidhuber · Neural Information Processing Systems · 2013
Local competition among neighboring neurons is common in biological neural networks (NNs). In this paper, we apply the concept to gradient-based, backprop-trained artificial multilayer NNs. NNs with competing linear units tend to outperform those with non-competing nonlinear units, and avoid catastrophic forgetting when training sets change over time.