Reproduce “Generating functionals for computational intelligence: the Fisher information as an objective function for self-limiting Hebbian learning rules” 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

Rodrigo Echeveste, Claudius Gros

The authors propose a rate-neuron rule combining Hebbian updates with a multiplicative brake that reverses potentiation when drive grows too large. Motivated by Fisher information under a local-synapse approximation, it limits growth in simulations and tends to select lower-excess-kurtosis inputs of equal variance. In 1,000 runs, it selected lower-kurtosis inputs 88.8% of the time, a simulation result, not biological evidence.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.