Panda or Gibbon? 2024
Think you can spot the difference? The AI can't. Panda or Gibbon? A Beginner's Introduction to Adversarial Attacks is an interactive guide to how small changes in an image can mislead a machine learning model. I built it with D3.js and Idyll to introduce adversarial attacks through examples you can explore yourself. The guide focuses on the Fast Gradient Sign Method (FGSM), letting you compare original and altered images and see how two versions of ResNet-34 respond: one with standard training and one trained to resist attacks.
Recognition & Outreach
- Accepted to and presented at the 7th VISxAI Workshop at IEEE VIS 2024. Explore the workshop program
Core Features
- Compare original images with versions altered by an FGSM attack.
- See how standard and adversarially trained models respond differently.
- Explore individual predictions and the patterns shown in embedding views.
Presentation at the VISxAI Workshop
Video of my presentation at the 7th VISxAI workshop (starts from 1:30:44)
Links
Interactive explainable
Explore the live VISxAI demo.
Video demo
Watch the walkthrough on YouTube.
Skills
- Python
- PyTorch
- t-SNE
- Machine Learning
- Computer Vision
- Adversarial Machine Learning
- XAI Visualization
- D3.js
- Idyll-lang
Keywords
- Adversarial Machine Learning
- FGSM Attack
- Adversarial Attack
- Image Classification
- Visualization
- ResNet
- Model Robustness
Authors
Yuzhe You, Jian Zhao