A Theory of Non-acyclic Generative Flow Networks
Leo Brunswic, Yinchuan Li, Yushun Xu, Yijun Feng, Shangling Jui, Lizhuang Ma
Abstract
GFlowNets is a novel flow-based method for learning a stochastic policy to generate objects via a sequence of actions and with probability proportional to a given positive reward. We contribute to relaxing hypotheses limiting the application range of GFlowNets, in particular: acyclicity (or lack thereof). To this end, we extend the theory of GFlowNets on measurable spaces which includes continuous state spaces without cycle restrictions, and provide a generalization of cycles in this generalized context. We show that losses used so far push flows to get stuck into cycles and we define a family of losses solving this issue. Experiments on graphs and continuous tasks validate those principles.
BibTeX
@article{Brunswic_Li_Xu_Feng_Jui_Ma_2024, title={A Theory of Non-acyclic Generative Flow Networks}, volume={38}, url={https://ojs.aaai.org/index.php/AAAI/article/view/28989}, DOI={10.1609/aaai.v38i10.28989}, abstractNote={GFlowNets is a novel flow-based method for learning a stochastic policy to generate objects via a sequence of actions and with probability proportional to a given positive reward. We contribute to relaxing hypotheses limiting the application range of GFlowNets, in particular: acyclicity (or lack thereof). To this end, we extend the theory of GFlowNets on measurable spaces which includes continuous state spaces without cycle restrictions, and provide a generalization of cycles in this generalized context. We show that losses used so far push flows to get stuck into cycles and we define a family of losses solving this issue. Experiments on graphs and continuous tasks validate those principles.}, number={10}, journal={Proceedings of the AAAI Conference on Artificial Intelligence}, author={Brunswic, Leo and Li, Yinchuan and Xu, Yushun and Feng, Yijun and Jui, Shangling and Ma, Lizhuang}, year={2024}, month={Mar.}, pages={11124-11131} }