# Bayesian Program Learning by Decompiling Amortized Knowledge

arXiv: https://arxiv.org/abs/2306.07856 · HTML: https://mlsanity.com/paper/2306.07856/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 15 of its 39 references that are in the map.
- First posted on arXiv 13 Jun 2023.

## Builds on (most cited first, 15 of 15)

- [Overcoming catastrophic forgetting in neural networks](https://mlsanity.com/paper/1612.00796/) (2016)
- [DreamCoder: Growing generalizable, interpretable knowledge with wake-sleep Bayesian program learning](https://mlsanity.com/paper/2006.08381/) (2020)
- [DeepCoder: Learning to Write Programs](https://mlsanity.com/paper/1611.01989/) (2016)
- [RobustFill: Neural Program Learning under Noisy I/O](https://mlsanity.com/paper/1703.07469/) (2017)
- [Reweighted Wake-Sleep](https://mlsanity.com/paper/1406.2751/) (2014)
- [Top-Down Synthesis for Library Learning](https://mlsanity.com/paper/2211.16605/) (2022)
- [Leveraging Language to Learn Program Abstractions and Search Heuristics](https://mlsanity.com/paper/2106.11053/) (2021)
- [Neural-Guided Deductive Search for Real-Time Program Synthesis from Examples](https://mlsanity.com/paper/1804.01186/) (2018)
- [Program Synthesis and Semantic Parsing with Learned Code Idioms](https://mlsanity.com/paper/1906.10816/) (2019)
- [Inference Compilation and Universal Probabilistic Programming](https://mlsanity.com/paper/1610.09900/) (2016)
- [babble: Learning Better Abstractions with E-Graphs and Anti-Unification](https://mlsanity.com/paper/2212.04596/) (2022)
- [Beyond imitation: Zero-shot task transfer on robots by learning concepts as cognitive programs](https://mlsanity.com/paper/1812.02788/) (2018)
- [Learning Programmatic Idioms for Scalable Semantic Parsing](https://mlsanity.com/paper/1904.09086/) (2019)
- [Learning to learn generative programs with Memoised Wake-Sleep](https://mlsanity.com/paper/2007.03132/) (2020)
- [SyGuS-Comp 2017: Results and Analysis](https://mlsanity.com/paper/1711.11438/) (2017)

## Built on by (most cited first, 2 of 2)

- [Program Synthesis via Test-Time Transduction](https://mlsanity.com/paper/2509.17393/) (2025)
- [Common Benchmarks Undervalue the Generalization Power of Programmatic Policies](https://mlsanity.com/paper/2506.14162/) (2025)

## Abstract (arXiv)

DreamCoder is an inductive program synthesis system that, whilst solving problems, learns to simplify search in an iterative wake-sleep procedure. The cost of search is amortized by training a neural search policy, reducing search breadth and effectively "compiling" useful information to compose program solutions across tasks. Additionally, a library of program components is learnt to compress and express discovered solutions in fewer components, reducing search depth. We present a novel approach for library learning that directly leverages the neural search policy, effectively "decompiling" its amortized knowledge to extract relevant program components. This provides stronger amortized inference: the amortized knowledge learnt to reduce search breadth is now also used to reduce search depth. We integrate our approach with DreamCoder and demonstrate faster domain proficiency with improved generalization on a range of domains, particularly when fewer example solutions are available.