mlsanityData to 7 Oct 2026

Pessimistic Backward Policy for GFlowNets

arXiv 2405.16012 ↗First posted 25 May 2024cs.LGcited by 13

Not on the map. Why?

Abstract (arXiv)

This paper studies Generative Flow Networks (GFlowNets), which learn to sample objects proportionally to a given reward function through the trajectory of state transitions. In this work, we observe that GFlowNets tend to under-exploit the high-reward objects due to training on insufficient number of trajectories, which may lead to a large gap between the estimated flow and the (known) reward value. In response to this challenge, we propose a pessimistic backward policy for GFlowNets (PBP-GFN), which maximizes the observed flow to align closely with the true reward for the object. We extensively evaluate PBP-GFN across eight benchmarks, including hyper-grid environment, bag generation, structured set generation, molecular generation, and four RNA sequence generation tasks. In particular, PBP-GFN enhances the discovery of high-reward objects, maintains the diversity of the objects, and consistently outperforms existing methods.

Key facts. Cited by 13 papers in the map (data through 7 Oct 2026). Builds on 31 of its 41 references that are in the map. First posted on arXiv 25 May 2024.

Builds on

31 of 41 references.

References · 31

  1. 2018How Powerful are Graph Neural Networks?cited by 4,355
  2. 2016Asynchronous Methods for Deep Reinforcement Learningcited by 2,218
  3. 2021Flow Network based Generative Models for Non-Iterative Diverse Candidate Generationcited by 384
  4. 2021GFlowNet Foundationscited by 286
  5. 2022Trajectory balance: Improved credit assignment in GFlowNetscited by 244
  6. 2022Biological Sequence Design with GFlowNetscited by 172
  7. 2022Bayesian Structure Learning with Generative Flow Networkscited by 153
  8. 2022Learning GFlowNets from partial episodes for improved convergence and stabilitycited by 129
  9. 2022Generative Flow Networks for Discrete Probabilistic Modelingcited by 113
  10. 2022GFlowNets and variational inferencecited by 103
  11. 2022Multi-Objective GFlowNetscited by 100
  12. 2023A theory of continuous generative flow networkscited by 98
19 more
  1. 2022Design-Bench: Benchmarks for Data-Driven Offline Model-Based Optimizationcited by 96
  2. 2023Amortizing intractable inference in large language modelscited by 94
  3. 2023Better Training of GFlowNets with Local Credit and Incomplete Trajectoriescited by 83
  4. 2023Towards Understanding and Improving GFlowNet Trainingcited by 74
  5. 2023Robust Scheduling with GFlowNetscited by 67
  6. 2020AdaLead: A simple and robust adaptive greedy search algorithm for sequence designcited by 63
  7. 2023Generative Flow Networks as Entropy-Regularized RLcited by 60
  8. 2023Diffusion Generative Flow Samplers: Improving learning signals through partial trajectory optimizationcited by 57
  9. 2024Ant Colony Sampling with GFlowNets for Combinatorial Optimizationcited by 53
  10. 2022Generative Augmented Flow Networkscited by 48
  11. 2023Sample-efficient Multi-objective Molecular Optimization with GFlowNetscited by 48
  12. 2023Local Search GFlowNetscited by 45
  13. 2024Discrete Probabilistic Inference as Control in Multi-path Environmentscited by 40
  14. 2023Thompson sampling for improved exploration in GFlowNetscited by 31
  15. 2023Maximum entropy GFlowNets with soft Q-learningcited by 30
  16. 2023Learning to Scale Logits for Temperature-Conditional GFlowNetscited by 26
  17. 2023Learning Energy Decompositions for Partial Inference of GFlowNetscited by 24
  18. 2024QGFN: Controllable Greediness with Action Valuescited by 18
  19. 2023Order-Preserving GFlowNetscited by 13

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 3; 2025: 4; 2026: 6.

13 papers.

Sort:

By area:

  1. 2024Adaptive teachers for amortized samplerscited by 23
  2. 2024Optimizing Backward Policies in GFlowNets via Trajectory Likelihood Maximizationcited by 12
  3. 2024Beyond Squared Error: Exploring Loss Design for Enhanced Training of Generative Flow Networkscited by 9
  4. 2025Revisiting Non-Acyclic GFlowNets in Discrete Environmentscited by 8
  5. 2025Adaptive Destruction Processes for Diffusion Samplerscited by 6
  6. 2025Scalable and Cost-Efficient de Novo Template-Based Molecular Generationcited by 5
  7. 2026Learning Shortest Paths with Generative Flow Networkscited by 2
  8. 2026Evaluating GFlowNet from partial episodes for stable and flexible policy-based trainingcited by 1
  9. 2025Beyond the Proxy: Trajectory-Distilled Guidance for Offline GFlowNet Trainingcited by 0
  10. 2026Flow of Spans: Generalizing Language Models to Dynamic Span-Vocabulary via GFlowNetscited by 0
  11. 2026A Systematic Evaluation of Co-folding Model Representations for Small-Molecule Learningcited by 0
  12. 2026Stable GFlowNets with TV Monitoring and Probabilistic Guaranteescited by 0
1 more
  1. 2026Your GFlowNet Secretly Learns an Optimal Transport Plancited by 0