Cooperative graphical models
We study a rich family of distributions that capture variable interactions significantly more expressive than those representable with low-treewidth or pairwise graphical models, or log-supermodular models. We call these cooperative graphical models. Yet, this family retains structure, which we care...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Morgan Kaufmann Publishers
2021
|
| Online Access: | https://hdl.handle.net/1721.1/129327 |
| _version_ | 1826194485926690816 |
|---|---|
| author | Jegelka, Stefanie Sabrina |
| author2 | Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science |
| author_facet | Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Jegelka, Stefanie Sabrina |
| author_sort | Jegelka, Stefanie Sabrina |
| collection | MIT |
| description | We study a rich family of distributions that capture variable interactions significantly more expressive than those representable with low-treewidth or pairwise graphical models, or log-supermodular models. We call these cooperative graphical models. Yet, this family retains structure, which we carefully exploit for efficient inference techniques. Our algorithms combine the polyhedral structure of submodular functions in new ways with variational inference methods to obtain both lower and upper bounds on the partition function. While our fully convex upper bound is minimized as an SDP or via tree-reweighted belief propagation, our lower bound is tightened via belief propagation or mean-field algorithms. The resulting algorithms are easy to implement and, as our experiments show, effectively obtain good bounds and marginals for synthetic and real-world examples. |
| first_indexed | 2024-09-23T09:56:30Z |
| format | Article |
| id | mit-1721.1/129327 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T09:56:30Z |
| publishDate | 2021 |
| publisher | Morgan Kaufmann Publishers |
| record_format | dspace |
| spelling | mit-1721.1/1293272022-09-30T17:53:26Z Cooperative graphical models Jegelka, Stefanie Sabrina Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science We study a rich family of distributions that capture variable interactions significantly more expressive than those representable with low-treewidth or pairwise graphical models, or log-supermodular models. We call these cooperative graphical models. Yet, this family retains structure, which we carefully exploit for efficient inference techniques. Our algorithms combine the polyhedral structure of submodular functions in new ways with variational inference methods to obtain both lower and upper bounds on the partition function. While our fully convex upper bound is minimized as an SDP or via tree-reweighted belief propagation, our lower bound is tightened via belief propagation or mean-field algorithms. The resulting algorithms are easy to implement and, as our experiments show, effectively obtain good bounds and marginals for synthetic and real-world examples. Swiss National Foundation for the Promotion of Scientific Research (Grant CRSII2147633) European Research Council (Grant StG 307036) National Science Foundation (U.S.). Career Grant (1553284) 2021-01-07T17:15:59Z 2021-01-07T17:15:59Z 2016-12 2020-12-21T19:11:06Z Article http://purl.org/eprint/type/ConferencePaper 1049-5258 https://hdl.handle.net/1721.1/129327 Djolonga, Josip et al. “Cooperative graphical models.” Advances in Neural Information Processing Systems, 29 ( December 2016 © 2016 The Author(s) en https://papers.nips.cc/paper/2016/hash/8f85517967795eeef66c225f7883bdcb-Abstract.html Advances in Neural Information Processing Systems Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf Morgan Kaufmann Publishers Neural Information Processing Systems (NIPS) |
| spellingShingle | Jegelka, Stefanie Sabrina Cooperative graphical models |
| title | Cooperative graphical models |
| title_full | Cooperative graphical models |
| title_fullStr | Cooperative graphical models |
| title_full_unstemmed | Cooperative graphical models |
| title_short | Cooperative graphical models |
| title_sort | cooperative graphical models |
| url | https://hdl.handle.net/1721.1/129327 |
| work_keys_str_mv | AT jegelkastefaniesabrina cooperativegraphicalmodels |