# Learning Neurosymbolic Generative Models via Program Synthesis

arXiv: https://arxiv.org/abs/1901.08565 · HTML: https://mlsanity.com/paper/1901.08565/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 24 Jan 2019.

## Built on by (most cited first, 11 of 11)

- [Neuro-Symbolic Artificial Intelligence: Current Trends](https://mlsanity.com/paper/2105.05330/) (2021)
- [Towards Data-and Knowledge-Driven Artificial Intelligence: A Survey on Neuro-Symbolic Computing](https://mlsanity.com/paper/2210.15889/) (2022)
- [Hierarchical Motion Understanding via Motion Programs](https://mlsanity.com/paper/2104.11216/) (2021)
- [Neurosymbolic Transformers for Multi-Agent Communication](https://mlsanity.com/paper/2101.03238/) (2021)
- [Program Synthesis Guided Reinforcement Learning for Partially Observed Environments](https://mlsanity.com/paper/2102.11137/) (2021)
- [Programmatic Concept Learning for Human Motion Description and Synthesis](https://mlsanity.com/paper/2206.13502/) (2022)
- [ImageEye: Batch Image Processing Using Program Synthesis](https://mlsanity.com/paper/2304.03253/) (2023)
- [Interpreting Expert Annotation Differences in Animal Behavior](https://mlsanity.com/paper/2106.06114/) (2021)
- [Leveraging Systematic Knowledge of 2D Transformations](https://mlsanity.com/paper/2206.00893/) (2022)
- [Bridging Neural and Symbolic Representations with Transitional Dictionary Learning](https://mlsanity.com/paper/2308.02000/) (2023)
- [Synthesizing Machine Learning Programs with PAC Guarantees via Statistical Sketching](https://mlsanity.com/paper/2110.05390/) (2021)

## Abstract (arXiv)

Significant strides have been made toward designing better generative models in recent years. Despite this progress, however, state-of-the-art approaches are still largely unable to capture complex global structure in data. For example, images of buildings typically contain spatial patterns such as windows repeating at regular intervals; state-of-the-art generative methods can't easily reproduce these structures. We propose to address this problem by incorporating programs representing global structure into the generative model---e.g., a 2D for-loop may represent a configuration of windows. Furthermore, we propose a framework for learning these models by leveraging program synthesis to generate training data. On both synthetic and real-world data, we demonstrate that our approach is substantially better than the state-of-the-art at both generating and completing images that contain global structure.