Summary of When to Trust Your Data: Enhancing Dyna-style Model-based Reinforcement Learning with Data Filter, by Yansong Li et al.
When to Trust Your Data: Enhancing Dyna-Style Model-Based Reinforcement Learning With Data Filter
by Yansong Li, Zeyu Dong, Ertai Luo, Yu Wu, Shuo Wu, Shuo Han
First submitted to arxiv on: 16 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Systems and Control (eess.SY)
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 limitations of reinforcement learning algorithms by introducing an out-of-distribution data filter to enhance simulated data quality. The filter removes data that significantly diverges from real-world data, improving critic updates and overall policy optimization. This approach integrates with the model-based policy optimization algorithm and achieves better performance in fewer interactions with the environment. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Reinforcement learning algorithms can be divided into two classes: model-free and model-based. Dyna-style algorithms combine these approaches to accelerate training, but their efficiency is compromised when the estimated environmental model is inaccurate. To address this issue, researchers introduce an out-of-distribution data filter that removes simulated data that significantly diverges from real-world data. This technique enhances simulated data quality and improves policy optimization. |
Keywords
» Artificial intelligence » Optimization » Reinforcement learning
