{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/barrier-frank-wolfe-for-marginal-inference","title":"Barrier Frank-Wolfe for Marginal Inference","arxiv_id":"1511.02124","date":"2015-11-06","proceeding":"NeurIPS 2015 12","authors":["Rahul G. Krishnan","Simon Lacoste-Julien","David Sontag"],"abstract":"We introduce a globally-convergent algorithm for optimizing the\ntree-reweighted (TRW) variational objective over the marginal polytope. The\nalgorithm is based on the conditional gradient method (Frank-Wolfe) and moves\npseudomarginals within the marginal polytope through repeated maximum a\nposteriori (MAP) calls. This modular structure enables us to leverage black-box\nMAP solvers (both exact and approximate) for variational inference, and obtains\nmore accurate results than tree-reweighted algorithms that optimize over the\nlocal consistency relaxation. Theoretically, we bound the sub-optimality for\nthe proposed algorithm despite the TRW objective having unbounded gradients at\nthe boundary of the marginal polytope. Empirically, we demonstrate the\nincreased quality of results found by tightening the relaxation over the\nmarginal polytope as well as the spanning tree polytope on synthetic and\nreal-world instances.","url_abs":"http://arxiv.org/abs/1511.02124v2","url_pdf":"http://arxiv.org/pdf/1511.02124v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"barrier-frank-wolfe-for-marginal-inference","repo_url":"https://github.com/clinicalml/fw-inference","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"variational-inference","task_name":"Variational Inference"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1511.02124","atlas_url":"https://app.syntology.ai/?focus=1511.02124","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}