Reproduce “Heuristic-Based Weak Learning for Automated Decision-Making” 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
Ryan Steed, Benjamin Williams
The authors present a method for training preference models from people’s ranked decision rules rather than large sets of labeled examples. Researchers translate heuristics into functions that may abstain, combine their labels using popularity weights, and train a classifier. In a kidney-exchange task, weighted heuristic labels achieved 85.6% agreement with respondents, near the supervised baseline’s 86.0%, on existing survey data.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.