Summary of Learning Interpretable Classifiers For Pddl Planning, by Arnaud Lequen
Learning Interpretable Classifiers for PDDL Planning
by Arnaud Lequen
First submitted to arxiv on: 13 Oct 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
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 creating interpretable models that understand an agent’s behavior compared to other agents on similar planning tasks expressed in PDDL. The approach involves learning logical formulas from a small set of examples demonstrating how an agent solves small planning instances. These formulas are human-readable, provide partial descriptions of an agent’s policy, and generalize to unseen cases. However, the researchers find that learning these formulas is computationally intractable due to its NP-hard nature. To overcome this challenge, they propose a topology-guided compilation to MaxSAT, which enables the generation of various formulas. The results show that accurate and interesting formulas can be learned within a reasonable time frame. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper explores how to create models that understand an agent’s behavior compared to others on similar planning tasks. The researchers use logical formulas to describe an agent’s policy, making it easier for humans to understand. They find that creating these formulas is very difficult because it’s an extremely hard problem. To solve this challenge, they use a special technique called MaxSAT, which helps them generate many different formulas. The results show that they can create accurate and useful formulas in a reasonable amount of time. |
