Summary of Hierarchical Preference Optimization: Learning to Achieve Goals Via Feasible Subgoals Prediction, by Utsav Singh et al.
Hierarchical Preference Optimization: Learning to achieve goals via feasible subgoals prediction
by Utsav Singh, Souradip Chakraborty, Wesley A. Suttle, Brian M. Sadler, Anit Kumar Sahu, Mubarak Shah, Vinay P. Namboodiri, Amrit Singh Bedi
First submitted to arxiv on: 1 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 research introduces Hierarchical Preference Optimization (HPO), a novel approach to hierarchical reinforcement learning that addresses challenges in complex robotic control tasks. By combining maximum entropy reinforcement learning with token-level Direct Preference Optimization, HPO eliminates the need for pre-trained reference policies, which are often unavailable in challenging scenarios. The authors formulate HRL as a bi-level optimization problem and transform it into a primitive-regularized DPO formulation to ensure feasible subgoal generation and avoid degenerate solutions. Experimental results on robotic navigation and manipulation tasks demonstrate impressive performance of HPO, achieving up to 35% improvement over baselines. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper introduces Hierarchical Preference Optimization (HPO), a new way for robots to learn and control complex movements. Traditionally, robots need a “map” or reference policy to follow, but this can be hard to find in challenging situations. HPO fixes this by using two techniques together: maximum entropy reinforcement learning and token-level Direct Preference Optimization. This allows the robot to figure out what it needs to do without needing a pre-made plan. The researchers tested HPO on robots that needed to navigate and manipulate objects, showing that it performed much better than other methods. |
Keywords
» Artificial intelligence » Optimization » Reinforcement learning » Token
