Reproduce “Collaborative Filtering and the Missing at Random Assumption” 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
Benjamin Marlin, Richard S. Zemel, Sam Roweis, Malcolm Slaney
The authors test whether modeling rating selection improves recommendation on randomly collected song ratings. Their CPT-v model represents the probability that a rating is observed as depending on its value and models user preferences. On LaunchCast’s random-song test set, CPT-v’s MAE was 0.7148 versus 1.2126 for the baseline, using probabilities estimated from held-out survey users; learned probabilities performed less strongly.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.