mlsanityData to 7 Oct 2026

Reclaiming the Source of Programmatic Policies: Programmatic versus Latent Spaces

arXiv 2410.12166 ↗First posted 16 Oct 2024cs.LG cs.AIcited by 2

Not on the map. Why?

Abstract (arXiv)

Recent works have introduced LEAPS and HPRL, systems that learn latent spaces of domain-specific languages, which are used to define programmatic policies for partially observable Markov decision processes (POMDPs). These systems induce a latent space while optimizing losses such as the behavior loss, which aim to achieve locality in program behavior, meaning that vectors close in the latent space should correspond to similarly behaving programs. In this paper, we show that the programmatic space, induced by the domain-specific language and requiring no training, presents values for the behavior loss similar to those observed in latent spaces presented in previous work. Moreover, algorithms searching in the programmatic space significantly outperform those in LEAPS and HPRL. To explain our results, we measured the "friendliness" of the two spaces to local search algorithms. We discovered that algorithms are more likely to stop at local maxima when searching in the latent space than when searching in the programmatic space. This implies that the optimization topology of the programmatic space, induced by the reward function in conjunction with the neighborhood function, is more conducive to search than that of the latent space. This result provides an explanation for the superior performance in the programmatic space.

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 14 of its 28 references that are in the map. First posted on arXiv 16 Oct 2024.

Builds on

14 of 28 references.

References · 14

  1. 2018Programmatically Interpretable Reinforcement Learningcited by 192
  2. 2018Verifiable Reinforcement Learning via Policy Extractioncited by 176
  3. 2020DreamCoder: Growing generalizable, interpretable knowledge with wake-sleep Bayesian program learningcited by 159
  4. 2018Leveraging Grammar and Reinforcement Learning for Neural Program Synthesiscited by 83
  5. 2021Learning to Synthesize Programs as Interpretable and Generalizable Policiescited by 59
  6. 2021Leveraging Language to Learn Program Abstractions and Search Heuristicscited by 44
  7. 2020BUSTLE: Bottom-Up Program Synthesis Through Learning-Guided Explorationcited by 30
  8. 2023Hierarchical Programmatic Reinforcement Learning via Learning to Compose Programscited by 20
  9. 2020Just-in-Time Learning for Bottom-Up Enumerative Synthesiscited by 18
  10. 2022CrossBeam: Learning to Search in Bottom-Up Program Synthesiscited by 18
  11. 2022What can we Learn Even From the Weakest? Learning Sketches for Programmatic Strategiescited by 10
  12. 2023Choosing Well Your Opponents: How to Guide the Synthesis of Programmatic Strategiescited by 8
2 more
  1. 2023Program Synthesis with Best-First Bottom-Up Searchcited by 7
  2. 2023Synthesizing Programmatic Policies with Actor-Critic Algorithms and ReLU Networkscited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 1; 2026: 1.

2 papers.

  1. 2025InnateCoder: Learning Programmatic Options with Foundation Modelscited by 2
  2. 2026Scheduling That Speaks: An Interpretable Programmatic Reinforcement Learning Frameworkcited by 0