Summary of Learning to Infer Generative Template Programs For Visual Concepts, by R. Kenny Jones et al.
Learning to Infer Generative Template Programs for Visual Concepts
by R. Kenny Jones, Siddhartha Chaudhuri, Daniel Ritchie
First submitted to arxiv on: 20 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Graphics (cs.GR); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The neurosymbolic system learns to infer programs that capture visual concepts in a domain-general fashion, introducing Template Programs as programmatic expressions from a domain-specific language. This framework supports multiple concept-related tasks, including few-shot generation and co-segmentation through parsing. The learning paradigm trains networks to directly infer Template Programs from visual datasets containing concept groupings. Experiments across 2D layouts, Omniglot characters, and 3D shapes show that the method outperforms task-specific alternatives and performs competitively against domain-specific approaches for limited domains. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new computer system can learn to understand different visual concepts by finding patterns in a few examples. It uses “template programs” to describe these patterns, which helps it complete tasks like generating new images or identifying objects. The system is trained on datasets that show how different concepts are related. It performs well across various visual domains, such as layouts, characters, and shapes. |
Keywords
* Artificial intelligence * Few shot * Parsing
