Summary of Knee or Roc, by Veronica Wendt et al.
Knee or ROC
by Veronica Wendt, Byunggu Yu, Caleb Kelly, Junwhan Kim
First submitted to arxiv on: 14 Jan 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 The paper explores the limitations of self-attention transformers in image classification tasks with small datasets. Currently, tests focus on single-class image detection with known population representations. However, when dealing with multiple classes and unknown population representations, traditional accuracy calculations must adapt. The authors propose using the Receiver Operating Characteristic (ROC) accuracy threshold for multi-class input images, but acknowledge its limitations. Instead, they suggest calculating accuracy using the knee method to determine threshold values on an ad-hoc basis. The paper discusses results for a multi-class dataset created from CIFAR-10 images. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at how self-attention transformers work with small image datasets. Right now, tests are done with single classes and known population types. But what if we have many classes and don’t know the population types? That’s where this paper comes in. The authors say that using a special type of curve called a Receiver Operating Characteristic (ROC) can help with multi-class images. However, they also think that another way to figure out accuracy, called the knee method, might be better for some situations. |
Keywords
* Artificial intelligence * Image classification * Self attention
