Summary of Improve Cost Efficiency Of Active Learning Over Noisy Dataset, by Zan-kai Chong et al.
Improve Cost Efficiency of Active Learning over Noisy Dataset
by Zan-Kai Chong, Hiroyuki Ohsaki, Bryan Ng
First submitted to arxiv on: 2 Mar 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 paper proposes an active learning strategy for binary classification tasks where acquiring a positive instance incurs a significantly higher cost compared to negative instances. The authors consider cases where unlabeled data is abundant but labeling is expensive, such as in the financial industry where defaulted loans can lead to substantial losses. To address this issue, they propose a shifted normal distribution sampling function that samples from a wider range than typical uncertainty sampling. Their simulation shows that their proposed method limits noisy and positive label selection, achieving between 20% and 32% improved cost efficiency over different test datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this paper, researchers develop a new way to improve machine learning algorithms in situations where it’s very expensive to get accurate answers. They focus on problems where getting the wrong answer can be costly, like when lending money to people who might not pay back their loans. The team creates a special way of selecting which data points to label, which helps them save time and money while still getting good results. |
Keywords
* Artificial intelligence * Active learning * Classification * Machine learning
