Summary of Looking at Model Debiasing Through the Lens Of Anomaly Detection, by Vito Paolo Pastore et al.
Looking at Model Debiasing through the Lens of Anomaly Detection
by Vito Paolo Pastore, Massimiliano Ciranni, Davide Marinelli, Francesca Odone, Vittorio Murino
First submitted to arxiv on: 24 Jul 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: 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 presents a novel approach to reducing bias in deep neural networks by accurately predicting and identifying bias-conflicting samples. The authors focus on the realistic scenario where bias information is not available, showing that this can lead to compelling performance in bias mitigation. They propose an anomaly detection-based method for identifying bias-conflicting samples as outliers in the feature space of a biased model. This approach is coupled with bias-conflicting data upsampling and augmentation in a two-step strategy, achieving state-of-the-art performance on synthetic and real benchmark datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us make better artificial intelligence (AI) that isn’t biased against certain groups or things. When AI models are trained, they can learn to favor certain types of data over others, which is called bias. This makes the model not very good at making decisions when it’s given new, different information. The authors came up with a new way to find and fix this problem by looking for patterns in the data that don’t fit with what the AI has learned so far. They tested their idea on some sample datasets and found that it worked really well. |
Keywords
* Artificial intelligence * Anomaly detection
