Summary of On the Utility Of Probing Trajectories For Algorithm-selection, by Quentin Renau and Emma Hart
On the Utility of Probing Trajectories for Algorithm-Selection
by Quentin Renau, Emma Hart
First submitted to arxiv on: 23 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Neural and Evolutionary Computing (cs.NE)
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 novel approach to algorithm selection in machine learning by shifting the focus from instance-based features to algorithm-centric descriptions. The traditional method views instances as input data, assuming that similar instances elicit similar performance from algorithms. However, this perspective overlooks how an algorithm perceives similarity between instances. To address this limitation, the authors propose a new approach that uses short probing trajectories calculated by applying a solver to an instance for a brief period. This trajectory-based representation is demonstrated to provide comparable or better results than computationally expensive landscape-based feature-based approaches. Furthermore, projecting the trajectories into a 2D space reveals that functions that are similar from an algorithm’s perspective do not necessarily align with human categorization. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper shows us a new way of thinking about how algorithms pick which one to use. Right now, we think about it like this: if two problems look similar, the same algorithm will work well on both. But what if our algorithm thinks differently? To help us figure this out, the researchers came up with a new method that looks at how an algorithm works for just a short time on each problem. This helps us understand which algorithms are best suited for certain types of problems. It’s like looking at a map from an airplane – you can see patterns and relationships that aren’t clear when you’re on the ground. |
Keywords
* Artificial intelligence * Machine learning
