Kickback cuts Backprop's red-tape: Biologically plausible credit assignment in neural networks
This work addresses the credit assignment problem in neural networks, offering a biologically plausible alternative to backpropagation, though it appears incremental as it builds on existing methods without a paradigm shift.
The paper tackles the biological implausibility and computational complexity of error backpropagation by introducing Kickback, a simpler credit assignment algorithm for neural networks, and demonstrates that it matches Backprop's performance on real-world regression benchmarks.
Error backpropagation is an extremely effective algorithm for assigning credit in artificial neural networks. However, weight updates under Backprop depend on lengthy recursive computations and require separate output and error messages -- features not shared by biological neurons, that are perhaps unnecessary. In this paper, we revisit Backprop and the credit assignment problem. We first decompose Backprop into a collection of interacting learning algorithms; provide regret bounds on the performance of these sub-algorithms; and factorize Backprop's error signals. Using these results, we derive a new credit assignment algorithm for nonparametric regression, Kickback, that is significantly simpler than Backprop. Finally, we provide a sufficient condition for Kickback to follow error gradients, and show that Kickback matches Backprop's performance on real-world regression benchmarks.