Summary of Batch Transformer: Look For Attention in Batch, by Myung Beom Her et al.
Batch Transformer: Look for Attention in Batch
by Myung Beom Her, Jisu Jeong, Hojoon Song, Ji-Hyeong Han
First submitted to arxiv on: 5 Jul 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
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 paper proposes a novel approach to facial expression recognition (FER) by introducing a batch transformer (BT) that addresses uncertainties in FER images, such as occlusion, low resolution, and pose variation. The BT consists of two key modules: class batch attention (CBA), which prevents overfitting in noisy data and extracts trustworthy information from multiple images in a batch; and multi-level attention (MLA), which captures correlations between different levels to prevent overfitting specific features. The authors combine these proposals into a batch transformer network (BTN) that outperforms state-of-the-art methods on various FER benchmark datasets, demonstrating the promise of this approach for improving FER performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper tries to make facial expression recognition better by using a new way to look at many images together. This helps remove mistakes caused by things like blurry faces or different angles. They also use another trick to make sure it doesn’t just focus on one type of feature, but instead looks at how all the features work together. The result is a system that can recognize facial expressions better than what’s out there right now. |
Keywords
» Artificial intelligence » Attention » Overfitting » Transformer
