Reproduce “Neural Dual Bounds: Valid-by-Construction JGLP Warm-Starts for MAP and Constrained MAP” 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
Akshay Vyas, Shivvrat Arya, Brij Malhotra, Tahrima Rahman, Vibhav Gogate, Nicholas Ruozzi
Neural Dual Bounds predicts canceling score transfers between neighboring clusters, producing valid upper bounds for MAP inference. Its architecture guarantees validity, while learning affects tightness. On BN_12, after 10 JGLP iterations, neural initialization reduced the mean absolute dual gap from 0.525 ± 0.248 to 0.304 ± 0.178. The gap uses the join-graph LP optimum as reference; gains concentrate early.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.