Reproduce “A Novel Predictive-Coding-Inspired Variational RNN Model for Online Prediction and Recognition” 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
Ahmadreza Ahmadi, Jun Tani
PV-RNN predicts another robot’s movements from observed hand positions by using prediction errors to update sequence-specific internal vectors while keeping learned weights fixed. Its five-step robot test reports a mean squared error of 0.00473, versus 0.00943 for VRNN. In a separate generated-sequence test, an intermediate meta-prior improved primitive prediction, but robot testing used prerecorded data because updates exceeded real-time limits.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.