Dual Propagation: Accelerating Contrastive Hebbian Learning with Dyadic Neurons
Paper in proceeding, 2023

Activity difference based learning algorithms-such as contrastive Hebbian learning and equilibrium propagation-have been proposed as biologically plausible alternatives to error backpropagation. However, on traditional digital chips these algorithms suffer from having to solve a costly inference problem twice, making these approaches more than two orders of magnitude slower than back-propagation. In the analog realm equilibrium propagation may be promising for fast and energy efficient learning, but states still need to be inferred and stored twice. Inspired by lifted neural networks and compartmental neuron models we propose a simple energy based compartmental neuron model, termed dual propagation, in which each neuron is a dyad with two intrinsic states. At inference time these intrinsic states encode the error/activity duality through their difference and their mean respectively. The advantage of this method is that only a single inference phase is needed and that inference can be solved in layerwise closed-form. Experimentally we show on common computer vision datasets, including Imagenet32x32, that dual propagation performs equivalently to back-propagation both in terms of accuracy and runtime.

Author

Rasmus Kjær Høier

Chalmers, Electrical Engineering, Signal Processing and Biomedical Engineering

Dorian Staudt

Chalmers, Electrical Engineering, Signal Processing and Biomedical Engineering

Christopher Zach

Chalmers, Electrical Engineering, Signal Processing and Biomedical Engineering

Proceedings of Machine Learning Research

26403498 (eISSN)

Vol. 202 13141-13156

International Conference on Machine Learning
Hawaii, ,

Energy-based models for supervised deep neural networks and their applications

Chalmers AI Research Centre (CHAIR), 2020-10-01 -- 2025-09-30.

Chalmers AI Research Centre (CHAIR), 2020-10-01 -- .

Subject Categories

Neurosciences

Computer Vision and Robotics (Autonomous Systems)

More information

Latest update

10/31/2023