Summary of Discovering Mathematical Formulas From Data Via Gpt-guided Monte Carlo Tree Search, by Yanjie Li et al.
Discovering Mathematical Formulas from Data via GPT-guided Monte Carlo Tree Search
by Yanjie Li, Weijun Li, Lina Yu, Min Wu, Jingyi Liu, Wenqiang Li, Meilan Hao, Shu Wei, Yusong Deng
First submitted to arxiv on: 24 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 In this paper, researchers tackle the challenge of finding concise mathematical formulas that accurately describe relationships between variables and predicted values in data. This problem, known as symbolic regression, is NP-hard. A previous algorithm using Monte Carlo Tree Search (MCTS) achieved state-of-the-art results on various datasets but was hindered by its lack of guidance during search. To optimize efficiency and versatility, the authors introduce SR-GPT, an algorithm integrating MCTS with a Generative Pre-Trained Transformer (GPT). GPT guides MCTS, improving search efficiency, while MCTS refines GPT for more accurate guidance. The authors couple MCTS and GPT to optimize each other until target expressions are determined. They evaluate SR-GPT using 222 expressions from over 10 symbolic regression datasets, demonstrating its superiority in accurately recovering symbolic expressions with and without added noise. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us understand how machines can find simple mathematical formulas that describe relationships between variables and predicted values in data. This is important because it’s hard to do this accurately. The researchers use a new method called SR-GPT, which combines two techniques: Monte Carlo Tree Search (MCTS) and Generative Pre-Trained Transformer (GPT). GPT helps MCTS find the right formula faster, while MCTS helps make GPT better. They tested their method on many formulas from different datasets and found that it works much better than other methods. |
Keywords
* Artificial intelligence * Gpt * Regression * Transformer
