Reproduce “Stochastic Forward-Forward Learning through Representational Dimensionality Compression” 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
Zhichao Zhu, Yang Qi, Hengyuan Ma, Wenlian Lu, Jianfeng Feng
The authors replace negative examples in Forward-Forward learning with an effective-dimensionality objective. It minimizes response dimensionality across noisy views of each input while maximizing it across batches. On CIFAR-10, the method reaches 76.96 ± 0.73% validation accuracy, versus 82.50 ± 0.09% for backpropagation under the same architecture and training protocol in reported shallow-network experiments.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.