Direct Feedback Alignment Scales to Modern Deep Learning Tasks and Architectures

TL;DR

This paper demonstrates that Direct Feedback Alignment can effectively train modern deep learning architectures across various tasks, achieving performance close to traditional backpropagation without requiring weight symmetry or sequential updates.

Contribution

It provides the first comprehensive study showing DFA's scalability to diverse deep learning tasks and architectures beyond computer vision.

Findings

DFA trains state-of-the-art architectures effectively

Performance close to backpropagation on multiple tasks

Challenging tasks can be addressed without weight transport

Abstract

Despite being the workhorse of deep learning, the backpropagation algorithm is no panacea. It enforces sequential layer updates, thus preventing efficient parallelization of the training process. Furthermore, its biological plausibility is being challenged. Alternative schemes have been devised; yet, under the constraint of synaptic asymmetry, none have scaled to modern deep learning tasks and architectures. Here, we challenge this perspective, and study the applicability of Direct Feedback Alignment to neural view synthesis, recommender systems, geometric learning, and natural language processing. In contrast with previous studies limited to computer vision tasks, our findings show that it successfully trains a large range of state-of-the-art deep learning architectures, with performance close to fine-tuned backpropagation. At variance with common beliefs, our work supports that challenging tasks can be tackled in the absence of weight transport.