Summary of Breaking the Barrier: Enhanced Utility and Robustness in Smoothed Drl Agents, by Chung-en Sun et al.
Breaking the Barrier: Enhanced Utility and Robustness in Smoothed DRL Agents
by Chung-En Sun, Sicun Gao, Tsui-Wei Weng
First submitted to arxiv on: 26 Jun 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 Medium Difficulty summary: Robustness remains a critical concern in deep reinforcement learning (DRL), where randomized smoothing has emerged as a key technique for enhancing this attribute. The study introduces novel algorithms, S-DQN and S-PPO, to train effective smoothed robust DRL agents, demonstrating remarkable improvements in clean rewards, empirical robustness, and robustness guarantee across standard RL benchmarks. Notably, these agents significantly outperform existing smoothed agents by an average factor of 2.16 times under the strongest attack and surpass previous robustly-trained agents by an average factor of 2.13 times. This represents a significant leap forward in the field. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Low Difficulty summary: Researchers have been trying to make AI systems more reliable and resistant to attacks. They’ve developed a new technique called randomized smoothing, which helps improve this reliability. The study presents two new algorithms, S-DQN and S-PPO, that can train these reliable AI systems much better than before. These algorithms do a great job of getting rewards in ideal situations (clean rewards) and are more resistant to attacks (robustness). They even outperform other robustly-trained agents. |
Keywords
* Artificial intelligence * Reinforcement learning
