Reproduce “How Biological Synapses Self-Assemble Gradient Learning” 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
Qianli Liao, Liu Ziyin, Yulu Gan, Brian Cheung, Mark T. Harnett, Tomaso Poggio
The authors introduce Self-Assembling Learning (SAL), in which local synaptic rules organize initially random connections between two interacting streams into a learning circuit called the Self-Assembling Motif (SAM). On CIFAR-10, SAM reaches 51.91 ± 0.21% accuracy, versus 50.36 ± 0.16% for SGD and 48.16 ± 0.09% for Kolen–Pollack. Gradient-like learning is conditional on pathway alignment; the two-stream scaffold is assumed.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.