Multi-Objective GFlowNets

Moksh Jain, Sharath Chandra Raparthy, Alex Hernández-Garcı́a, Jarrid Rector-Brooks, Yoshua Bengio, Santiago Miret, Emmanuel Bengio
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:14631-14653, 2023.

Abstract

We study the problem of generating diverse candidates in the context of Multi-Objective Optimization. In many applications of machine learning such as drug discovery and material design, the goal is to generate candidates which simultaneously optimize a set of potentially conflicting objectives. Moreover, these objectives are often imperfect evaluations of some underlying property of interest, making it important to generate diverse candidates to have multiple options for expensive downstream evaluations. We propose Multi-Objective GFlowNets (MOGFNs), a novel method for generating diverse Pareto optimal solutions, based on GFlowNets. We introduce two variants of MOGFNs: MOGFN-PC, which models a family of independent sub-problems defined by a scalarization function, with reward-conditional GFlowNets, and MOGFN-AL, which solves a sequence of sub-problems defined by an acquisition function in an active learning loop. Our experiments on wide variety of synthetic and benchmark tasks demonstrate advantages of the proposed methods in terms of the Pareto performance and importantly, improved candidate diversity, which is the main contribution of this work.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-jain23a, title = {Multi-Objective {GF}low{N}ets}, author = {Jain, Moksh and Raparthy, Sharath Chandra and Hern\'{a}ndez-Garc\'{\i}a, Alex and Rector-Brooks, Jarrid and Bengio, Yoshua and Miret, Santiago and Bengio, Emmanuel}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {14631--14653}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/jain23a/jain23a.pdf}, url = {https://proceedings.mlr.press/v202/jain23a.html}, abstract = {We study the problem of generating diverse candidates in the context of Multi-Objective Optimization. In many applications of machine learning such as drug discovery and material design, the goal is to generate candidates which simultaneously optimize a set of potentially conflicting objectives. Moreover, these objectives are often imperfect evaluations of some underlying property of interest, making it important to generate diverse candidates to have multiple options for expensive downstream evaluations. We propose Multi-Objective GFlowNets (MOGFNs), a novel method for generating diverse Pareto optimal solutions, based on GFlowNets. We introduce two variants of MOGFNs: MOGFN-PC, which models a family of independent sub-problems defined by a scalarization function, with reward-conditional GFlowNets, and MOGFN-AL, which solves a sequence of sub-problems defined by an acquisition function in an active learning loop. Our experiments on wide variety of synthetic and benchmark tasks demonstrate advantages of the proposed methods in terms of the Pareto performance and importantly, improved candidate diversity, which is the main contribution of this work.} }
Endnote
%0 Conference Paper %T Multi-Objective GFlowNets %A Moksh Jain %A Sharath Chandra Raparthy %A Alex Hernández-Garcı́a %A Jarrid Rector-Brooks %A Yoshua Bengio %A Santiago Miret %A Emmanuel Bengio %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-jain23a %I PMLR %P 14631--14653 %U https://proceedings.mlr.press/v202/jain23a.html %V 202 %X We study the problem of generating diverse candidates in the context of Multi-Objective Optimization. In many applications of machine learning such as drug discovery and material design, the goal is to generate candidates which simultaneously optimize a set of potentially conflicting objectives. Moreover, these objectives are often imperfect evaluations of some underlying property of interest, making it important to generate diverse candidates to have multiple options for expensive downstream evaluations. We propose Multi-Objective GFlowNets (MOGFNs), a novel method for generating diverse Pareto optimal solutions, based on GFlowNets. We introduce two variants of MOGFNs: MOGFN-PC, which models a family of independent sub-problems defined by a scalarization function, with reward-conditional GFlowNets, and MOGFN-AL, which solves a sequence of sub-problems defined by an acquisition function in an active learning loop. Our experiments on wide variety of synthetic and benchmark tasks demonstrate advantages of the proposed methods in terms of the Pareto performance and importantly, improved candidate diversity, which is the main contribution of this work.
APA
Jain, M., Raparthy, S.C., Hernández-Garcı́a, A., Rector-Brooks, J., Bengio, Y., Miret, S. & Bengio, E.. (2023). Multi-Objective GFlowNets. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:14631-14653 Available from https://proceedings.mlr.press/v202/jain23a.html.

Related Material