mlsanityData to 14 Oct 2026

Path-dependent Discrete Amortized Inference

arXiv 2608.08644 ↗First posted 9 Aug 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

We consider the problem of sampling compositional and discrete objects from a given unnormalized posterior distribution. Notably, recent studies have shown that this problem can be efficiently solved by learning a deterministic Markov Decision Process (MDP) that progressively builds each object in proportion to the posterior. In this work, however, we demonstrate that the Markovian assumption can both hamper signal propagation during training and catastrophically reduce the learned sampler's expressivity due to state aliasing. To address these issues, we propose lifting the MDP with a learnable latent dynamical system that allows the underlying policy to depend on the entire past trajectory---and not only on the current state. In view of this, we refer to the resulting method as path-dependent discrete amortized inference. Importantly, we provably extend existing learning algorithms for discrete amortized samplers to our setting. In experiments on standard benchmark problems, we also show that our approach often leads to faster learning convergence and improved state space exploration relatively to prior techniques.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 59 of its 104 references that are in the map. First posted on arXiv 9 Aug 2026.

Builds on

59 of 104 references.

References · 59

  1. 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,361
  2. 2014Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translationcited by 4,057
  3. 2016Variational Inference: A Review for Statisticianscited by 1,472
  4. 2021Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gaugescited by 1,167
  5. 2016The Concrete Distribution: A Continuous Relaxation of Discrete Random Variablescited by 1,148
  6. 2013Exact solutions to the nonlinear dynamics of learning in deep linear neural networkscited by 763
  7. 2021Linear Transformers Are Secretly Fast Weight Programmerscited by 502
  8. 2020Inductive Representation Learning on Temporal Graphscited by 462
  9. 2018Lipschitz regularity of deep neural networks: analysis and efficient estimationcited by 415
  10. 2021Flow Network based Generative Models for Non-Iterative Diverse Candidate Generationcited by 392
  11. 2017The Expressive Power of Neural Networks: A View from the Widthcited by 320
  12. 2018A Survey of Inverse Reinforcement Learning: Challenges, Methods and Progresscited by 298
38 more
  1. 2021GFlowNet Foundationscited by 290
  2. 2022Trajectory balance: Improved credit assignment in GFlowNetscited by 251
  3. 2017A Conceptual Introduction to Hamiltonian Monte Carlocited by 209
  4. 2021Inductive Representation Learning in Temporal Networks via Causal Anonymous Walkscited by 193
  5. 2022Biological Sequence Design with GFlowNetscited by 174
  6. 2022Bayesian Structure Learning with Generative Flow Networkscited by 153
  7. 2017Underdamped Langevin MCMC: A non-asymptotic analysiscited by 133
  8. 2022Learning GFlowNets from partial episodes for improved convergence and stabilitycited by 132
  9. 2017Approximating Continuous Functions by ReLU Nets of Minimal Widthcited by 122
  10. 2021Recurrent Model-Free RL Can Be a Strong Baseline for Many POMDPscited by 122
  11. 2022GFlowNets and variational inferencecited by 106
  12. 2023A theory of continuous generative flow networkscited by 101
  13. 2022Design-Bench: Benchmarks for Data-Driven Offline Model-Based Optimizationcited by 97
  14. 2023Amortizing intractable inference in large language modelscited by 97
  15. 2021The Logic of Graph Neural Networkscited by 85
  16. 2023Better Training of GFlowNets with Local Credit and Incomplete Trajectoriescited by 85
  17. 2020VarGrad: A Low-Variance Gradient Estimator for Variational Inferencecited by 75
  18. 2023Towards Understanding and Improving GFlowNet Trainingcited by 75
  19. 2024Amortizing intractable inference in diffusion models for vision, language, and controlcited by 72
  20. 2023Robust Scheduling with GFlowNetscited by 69
  21. 2017Deep linear neural networks with arbitrary loss: All local minima are globalcited by 65
  22. 2023Generative Flow Networks as Entropy-Regularized RLcited by 64
  23. 2017QMDP-Net: Deep Learning for Planning under Partial Observabilitycited by 51
  24. 2022Generative Augmented Flow Networkscited by 50
  25. 2023Joint Bayesian Inference of Graphical Structure and Parameters with a Single Generative Flow Networkcited by 50
  26. 2024Discrete Probabilistic Inference as Control in Multi-path Environmentscited by 44
  27. 2022A Modern Self-Referential Weight Matrix That Learns to Modify Itselfcited by 41
  28. 2023Let the Flows Tell: Solving Graph Combinatorial Optimization Problems with GFlowNetscited by 40
  29. 2023Compositional Sculpting of Iterative Generative Processescited by 27
  30. 2023Learning to Scale Logits for Temperature-Conditional GFlowNetscited by 27
  31. 2023Learning Energy Decompositions for Partial Inference of GFlowNetscited by 24
  32. 2024Adaptive teachers for amortized samplerscited by 23
  33. 2023PhyloGFN: Phylogenetic inference with generative flow networkscited by 22
  34. 2023Towards Theoretical Understanding of Inverse Reinforcement Learningcited by 21
  35. 2023Generative Flow Networks: a Markov Chain Perspectivecited by 14
  36. 2023A Theory of Non-Acyclic Generative Flow Networkscited by 12
  37. 2024Beyond Squared Error: Exploring Loss Design for Enhanced Training of Generative Flow Networkscited by 11
  38. 2025Revisiting Non-Acyclic GFlowNets in Discrete Environmentscited by 10

50 most cited of 59 All 59 (JSON)

Built on by

Not cited yet.