Summary of From Biased Selective Labels to Pseudo-labels: An Expectation-maximization Framework For Learning From Biased Decisions, by Trenton Chang and Jenna Wiens
From Biased Selective Labels to Pseudo-Labels: An Expectation-Maximization Framework for Learning from Biased Decisions
by Trenton Chang, Jenna Wiens
First submitted to arxiv on: 27 Jun 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 This paper addresses the issue of selective labels in machine learning, where label observations are subject to a decision-making process. For instance, diagnoses that depend on laboratory tests can lead to biased labeling. The authors propose Disparate Censorship Expectation-Maximization (DCEM), an algorithm designed to learn in the presence of disparate censorship. They theoretically analyze how DCEM mitigates the effects of disparate censorship on model performance, achieving better bias mitigation without sacrificing discriminative performance compared to baselines. The paper also presents results on synthetic and clinical data, demonstrating the effectiveness of DCEM. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at a special kind of problem in machine learning where labels are not always clear-cut. Imagine doctors trying to diagnose illnesses based on test results – some patients might need multiple tests before they get diagnosed. This can lead to biased labeling, which is bad for AI models that learn from these labels. The authors came up with a new algorithm called DCEM that helps AI models deal with this kind of bias. They tested it and showed that it does a better job than other methods at reducing bias without sacrificing accuracy. |
Keywords
» Artificial intelligence » Machine learning
