Summary of Contrastive Approach to Prior Free Positive Unlabeled Learning, by Anish Acharya et al.
Contrastive Approach to Prior Free Positive Unlabeled Learning
by Anish Acharya, Sujay Sanghavi
First submitted to arxiv on: 8 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
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 a novel framework for Positive Unlabeled (PU) learning, which involves training a binary classifier using only a few labeled positive samples and a set of unlabeled samples that could be either positive or negative. The authors start by learning a feature space through pretext-invariant representation learning, followed by pseudo-labeling the unlabeled examples leveraging their concentration property in this learned embedding space. Compared to existing state-of-the-art PU learning methods, the proposed approach achieves superior performance on several standard benchmark datasets without requiring prior knowledge of class priors. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us learn how to identify things that are good or bad based on just a few examples where we know they’re good, and many more examples where we don’t know which ones are good or bad. The researchers found a way to make this process work better by first learning how to understand what makes something “good” or “bad”, then guessing whether the other things are good or bad based on that understanding. They tested their new method on several sets of data and found it worked much better than existing methods, even when there were very few examples where we knew they were good. |
Keywords
* Artificial intelligence * Embedding space * Representation learning
