Summary of Learning to Edit Visual Programs with Self-supervision, by R. Kenny Jones et al.
Learning to Edit Visual Programs with Self-Supervision
by R. Kenny Jones, Renhao Zhang, Aditya Ganeshan, Daniel Ritchie
First submitted to arxiv on: 4 Jun 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 This paper presents a system called an “edit network” designed to learn how to edit visual programs. The network takes in a complete input program and a target program, predicting local edit operations that improve the similarity between the two. To apply this approach to domains without annotated programs, the authors develop a self-supervised learning method combining the edit network with a one-shot program prediction model. By finetuning these networks together and using an inference procedure to evolve populations of visual programs, the system infers more accurate results than relying solely on the one-shot model. Experimental results across multiple domains demonstrate significant advantages of this editing-based approach over traditional methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about creating a computer program that can edit other programs to make them look more like what we want. The program, called an “edit network,” takes in two programs and figures out how to change the first one to make it similar to the second. To make this work for situations where we don’t have information on what the correct edited program should be, the authors came up with a way to train the program using only the one-shot model. By combining these two approaches, they were able to make more accurate predictions than just relying on the one-shot model. The results show that this new approach works well across different situations. |
Keywords
» Artificial intelligence » Inference » One shot » Self supervised
