Summary of Online Optimization Of Curriculum Learning Schedules Using Evolutionary Optimization, by Mohit Jiwatode et al.
Online Optimization of Curriculum Learning Schedules using Evolutionary Optimization
by Mohit Jiwatode, Leon Schlecht, Alexander Dockhorn
First submitted to arxiv on: 12 Aug 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- 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 RHEA CL combines Curriculum Learning with Rolling Horizon Evolutionary Algorithms to automatically produce effective curricula for reinforcement learning agents. The algorithm optimizes a population of curricula using an evolutionary approach, selecting the best-performing curriculum as the starting point for the next training epoch. Evaluations are conducted after every curriculum step in all environments. RHEA CL demonstrates adaptability and consistent improvement, particularly in early stages, while reaching a stable performance later that outperforms other curriculum learners. The algorithm is evaluated on Minigrid framework’s DoorKey and DynamicObstacles environments, showing performance improvements for the final RL agent at the cost of additional evaluation during training. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary RHEA CL helps train better reinforcement learning agents by creating good lesson plans automatically. It does this by using an evolutionary process to test different lesson plans and choose the best one. This approach is tested on two environments, DoorKey and DynamicObstacles, and shows that it can improve performance over time. |
Keywords
* Artificial intelligence * Curriculum learning * Reinforcement learning
