Summary of Just Cluster It: An Approach For Exploration in High-dimensions Using Clustering and Pre-trained Representations, by Stefan Sylvius Wagner and Stefan Harmeling
Just Cluster It: An Approach for Exploration in High-Dimensions using Clustering and Pre-Trained Representations
by Stefan Sylvius Wagner, Stefan Harmeling
First submitted to arxiv on: 5 Feb 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 The paper proposes a representation-centric perspective on exploration in reinforcement learning, viewing it as a density estimation problem. The authors investigate the effectiveness of clustering representations for exploration in 3-D environments, finding that even random features can be clustered effectively to count states, but pre-trained DINO representations are more effective when the environment becomes visually complex due to their pre-trained inductive biases. The method is evaluated on VizDoom and Habitat environments, showing that it surpasses other well-known exploration methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at how to make decisions in a game or virtual world by exploring different options. It’s like trying to find the best way through a maze! The authors come up with a new way of thinking about this problem, where they treat the decision-making process as a puzzle that needs to be solved. They test their idea on two different environments and show that it works really well. |
Keywords
* Artificial intelligence * Clustering * Density estimation * Reinforcement learning
