Reproduce “Products of Hidden Markov Models: It Takes N>1 to Tango” 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
Graham W Taylor, Geoffrey E. Hinton
The authors evaluate products of hidden Markov models for complex time series, using annealed importance sampling to estimate the partition function needed for likelihoods. Several smaller models contribute constraints to a shared sequence, distributing its hidden state. On held-out Ceará rainfall data, products generally outperform tested alternatives at comparable parameter counts on likelihood and classification accuracy, but not every metric.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.