Summary of Learning to Defer to a Population: a Meta-learning Approach, by Dharmesh Tailor et al.
Learning to Defer to a Population: A Meta-Learning Approach
by Dharmesh Tailor, Aditya Patra, Rajeev Verma, Putra Manggala, Eric Nalisnick
First submitted to arxiv on: 5 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Machine Learning (stat.ML)
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 learning to defer (L2D) framework allows autonomous systems to be safe and robust by allocating difficult decisions to a human expert. The existing work assumes that each expert is well-identified, but this paper alleviates this constraint by formulating an L2D system that can cope with never-before-seen experts at test-time using meta-learning. The framework quickly adapts its deferral policy given a small context set, and the model-based approach employs an attention mechanism to assess the expert’s abilities. The methods are validated on image recognition, traffic sign detection, and skin lesion diagnosis benchmarks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about making machines smarter by letting them work with humans better. Right now, machines can’t always make good decisions on their own, so they need help from experts. But what if the expert changes? That’s where this new framework comes in – it lets machines work with different experts without needing to be re-trained. It does this by learning how to adapt to new experts and make better decisions. |
Keywords
* Artificial intelligence * Attention * Meta learning
