Summary of S-e Pipeline: a Vision Transformer (vit) Based Resilient Classification Pipeline For Medical Imaging Against Adversarial Attacks, by Neha a S et al.
S-E Pipeline: A Vision Transformer (ViT) based Resilient Classification Pipeline for Medical Imaging Against Adversarial Attacks
by Neha A S, Vivek Chaturvedi, Muhammad Shafique
First submitted to arxiv on: 23 Jul 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
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 image classification pipeline called S-E Pipeline, which enhances the robustness of Vision Transformer (ViT) models against adversarial attacks. The pipeline employs multiple pre-processing steps, including segmentation and image enhancement techniques such as CLAHE, UM, and HFE, to identify critical features that remain intact even after perturbations. Experimental results demonstrate a significant reduction in the effect of adversarial attacks, with 72.22% improvement for ViT-b32 and 86.58% for ViT-l32. The proposed method is deployed on an NVIDIA Jetson Orin Nano board to showcase its practical application on resource-constrained devices. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps make medical imaging machines more accurate by creating a new way to process images called S-E Pipeline. It uses special techniques to highlight important parts of the image that can’t be changed even if someone tries to trick the machine with fake information. The new pipeline makes it much harder for bad guys to fool the machine, reducing errors by 72% and 86% in certain cases. The method is tested on a small computer chip called NVIDIA Jetson Orin Nano, showing how it can be used in real-life devices. |
Keywords
* Artificial intelligence * Image classification * Vision transformer * Vit
