Summary of Decision-focused Learning to Predict Action Costs For Planning, by Jayanta Mandi et al.
Decision-Focused Learning to Predict Action Costs for Planning
by Jayanta Mandi, Marco Foschini, Daniel Holler, Sylvie Thiebaux, Jorg Hoffmann, Tias Guns
First submitted to arxiv on: 13 Aug 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Robotics (cs.RO)
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 Decision-Focused Learning (DFL) is a successful approach in learning to predict parameters of combinatorial optimization problems, optimizing solution quality over prediction quality. In automated planning, DFL aims to learn action costs based on input features like weather forecasts, enhancing solution quality. This paper investigates implementing DFL for automated planning, addressing two main challenges: (1) gradient computation during learning, and (2) repeated planner calls hindering scalability. Novel methods are proposed to overcome these issues. The authors demonstrate that DFL yields significantly better plans than predicting action costs aimed at minimizing prediction error, and caching can temper computation requirements. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine trying to plan a trip without knowing how long it takes to get from one place to another, depending on the weather. This is a problem in many automated planning applications. A new approach called Decision-Focused Learning (DFL) tries to solve this by learning to predict these “action costs” based on input features like weather forecasts. In this paper, researchers investigate how to use DFL for automated planning, dealing with two main challenges: one about computing gradients during training and another about speeding up the process. They propose new methods to overcome these issues and show that DFL can produce better plans than just predicting action costs. |
Keywords
» Artificial intelligence » Optimization
