Reproduce “Teaching signal synchronization in deep neural networks with prospective neurons” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Nicolas Zucchet, Qianqian Feng, Axel Laborieux, Friedemann Zenke, Walter Senn, João Sacramento
The authors introduce prospective neurons that use estimated input changes to align learning signals with activity despite delayed neural responses. Slow, non-prospective activity still stores memories. In a simulated teacher–student task, prospective backpropagation achieves 1.13 × 10⁻³ test loss, close to instantaneous backpropagation’s 1.15 × 10⁻³ and far below leaky backpropagation’s 1.32 × 10¹; biological efficacy remains untested.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.