I am Sayak (সায়ক) Paul. This is my personal website and it includes my blog posts, coordinates, interviews, and stuff like that. Know more about me from here.
This site is built with fastpages, an easy to use blogging platform with extra features for Jupyter Notebooks.
- New blog post: Learning to Resize in Computer Vision.
- Open-sourced my implementation of Barlow Twins here. Barlow Twins introduces a loss function that inherently avoids representation collapse when training vision models with self-supervision.
- Open-sourced my implementation of PAWS here. PAWS introduces a way to incorporate a small fraction of labeled samples with unlabeled ones during the pre-training of vision models.
- Open-sourced my implementation of Denoised Smoothing here. It provides provable robustness for pre-trained image classification models (including cloud APIs) against L2 attacks.
- Our (with Pin-Yu Chen and Dipanjan Sarkar) tutorial on Practical Adversarial Robustness in Deep Learning: Problems and Solutions has been accepted for CVPR 2021. Details are available on this website.