Sivapriya Vellaichamy's Projects
Adversarial Robustness Toolbox (ART) - Python Library for Machine Learning Security - Evasion, Poisoning, Extraction, Inference - Red and Blue Teams
With applications in the field of archaeology, we model the process of restoration of Heritage Structures by using geometric similarities to reassemble fragments to their parent objects. We use Douglas Peucker algorithm for contour extraction and a modified version of Smith-Waterman Algorithm for pairwise matching
Exploration of the use of different type of attribution algorithms - both gradient and perturbation - for images in PyTorch: Saliency Maps, GradCAM, Fooling Images, Class Visualization.
The purpose of our project is to tackle this problem by evaluating data related to prevalence of mental health disorders in the IT workplace and generate a profile for adults most prone to depression. This will help understand the risk factors that contribute to mental health disorders. The insights can also help guide company policies regarding mental health resources to protect their employees in the workplace.
Visual explanation for understanding features that influence popularity of pet images: Use existing Deep Neural architectures (like VGG, Alexnet) to determine popularity of pets (cats & dogs) with the application of transfer learning techniques and use saliency maps to visually explain the important features
We take an image and add the style of another reference style image to it and give it a new look. We do this experiment inspired from ”Image Style Transfer Using Convolutional Neural Networks” (Gatys et al CVPR 2015).
Config files for my GitHub profile.