Proteins, Particles, and Pseudo-Max-Marginals: A Submodular Approach

Jason Pacheco, Erik Sudderth
Proceedings of the 32nd International Conference on Machine Learning, PMLR 37:2200-2208, 2015.

Abstract

Variants of max-product (MP) belief propagation effectively find modes of many complex graphical models, but are limited to discrete distributions. Diverse particle max-product (D-PMP) robustly approximates max-product updates in continuous MRFs using stochastically sampled particles, but previous work was specialized to tree-structured models. Motivated by the challenging problem of protein side chain prediction, we extend D-PMP in several key ways to create a generic MAP inference algorithm for loopy models. We define a modified diverse particle selection objective that is provably submodular, leading to an efficient greedy algorithm with rigorous optimality guarantees, and corresponding max-marginal error bounds. We further incorporate tree-reweighted variants of the MP algorithm to allow provable verification of global MAP recovery in many models. Our general-purpose Matlab library is applicable to a wide range of pairwise graphical models, and we validate our approach using optical flow benchmarks. We further demonstrate superior side chain prediction accuracy compared to baseline algorithms from the state-of-the-art Rosetta package.

Cite this Paper

BibTeX
@InProceedings{pmlr-v37-pacheco15, title = {Proteins, Particles, and Pseudo-Max-Marginals: A Submodular Approach}, author = {Pacheco, Jason and Sudderth, Erik}, booktitle = {Proceedings of the 32nd International Conference on Machine Learning}, pages = {2200--2208}, year = {2015}, editor = {Bach, Francis and Blei, David}, volume = {37}, series = {Proceedings of Machine Learning Research}, address = {Lille, France}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v37/pacheco15.pdf}, url = {https://proceedings.mlr.press/v37/pacheco15.html}, abstract = {Variants of max-product (MP) belief propagation effectively find modes of many complex graphical models, but are limited to discrete distributions. Diverse particle max-product (D-PMP) robustly approximates max-product updates in continuous MRFs using stochastically sampled particles, but previous work was specialized to tree-structured models. Motivated by the challenging problem of protein side chain prediction, we extend D-PMP in several key ways to create a generic MAP inference algorithm for loopy models. We define a modified diverse particle selection objective that is provably submodular, leading to an efficient greedy algorithm with rigorous optimality guarantees, and corresponding max-marginal error bounds. We further incorporate tree-reweighted variants of the MP algorithm to allow provable verification of global MAP recovery in many models. Our general-purpose Matlab library is applicable to a wide range of pairwise graphical models, and we validate our approach using optical flow benchmarks. We further demonstrate superior side chain prediction accuracy compared to baseline algorithms from the state-of-the-art Rosetta package.} }
Endnote
%0 Conference Paper %T Proteins, Particles, and Pseudo-Max-Marginals: A Submodular Approach %A Jason Pacheco %A Erik Sudderth %B Proceedings of the 32nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2015 %E Francis Bach %E David Blei %F pmlr-v37-pacheco15 %I PMLR %P 2200--2208 %U https://proceedings.mlr.press/v37/pacheco15.html %V 37 %X Variants of max-product (MP) belief propagation effectively find modes of many complex graphical models, but are limited to discrete distributions. Diverse particle max-product (D-PMP) robustly approximates max-product updates in continuous MRFs using stochastically sampled particles, but previous work was specialized to tree-structured models. Motivated by the challenging problem of protein side chain prediction, we extend D-PMP in several key ways to create a generic MAP inference algorithm for loopy models. We define a modified diverse particle selection objective that is provably submodular, leading to an efficient greedy algorithm with rigorous optimality guarantees, and corresponding max-marginal error bounds. We further incorporate tree-reweighted variants of the MP algorithm to allow provable verification of global MAP recovery in many models. Our general-purpose Matlab library is applicable to a wide range of pairwise graphical models, and we validate our approach using optical flow benchmarks. We further demonstrate superior side chain prediction accuracy compared to baseline algorithms from the state-of-the-art Rosetta package.
RIS
TY - CPAPER TI - Proteins, Particles, and Pseudo-Max-Marginals: A Submodular Approach AU - Jason Pacheco AU - Erik Sudderth BT - Proceedings of the 32nd International Conference on Machine Learning DA - 2015/06/01 ED - Francis Bach ED - David Blei ID - pmlr-v37-pacheco15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 37 SP - 2200 EP - 2208 L1 - http://proceedings.mlr.press/v37/pacheco15.pdf UR - https://proceedings.mlr.press/v37/pacheco15.html AB - Variants of max-product (MP) belief propagation effectively find modes of many complex graphical models, but are limited to discrete distributions. Diverse particle max-product (D-PMP) robustly approximates max-product updates in continuous MRFs using stochastically sampled particles, but previous work was specialized to tree-structured models. Motivated by the challenging problem of protein side chain prediction, we extend D-PMP in several key ways to create a generic MAP inference algorithm for loopy models. We define a modified diverse particle selection objective that is provably submodular, leading to an efficient greedy algorithm with rigorous optimality guarantees, and corresponding max-marginal error bounds. We further incorporate tree-reweighted variants of the MP algorithm to allow provable verification of global MAP recovery in many models. Our general-purpose Matlab library is applicable to a wide range of pairwise graphical models, and we validate our approach using optical flow benchmarks. We further demonstrate superior side chain prediction accuracy compared to baseline algorithms from the state-of-the-art Rosetta package. ER -
APA
Pacheco, J. & Sudderth, E.. (2015). Proteins, Particles, and Pseudo-Max-Marginals: A Submodular Approach. Proceedings of the 32nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 37:2200-2208 Available from https://proceedings.mlr.press/v37/pacheco15.html.

Related Material