Reproduce “Conceptors: an easy introduction” 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
Herbert Jaeger
The authors show how conceptors let one recurrent neural network replay human movements by filtering its activity with pattern-specific state-space signatures. In simulations, a 1,000-neuron network stored 15 motion patterns as 61-dimensional signals, while a separate 100-neuron network nearly exhausted its memory quota at 15 synthetic patterns and failed on the 16th. These results concern artificial RNNs, not brains.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.