Vikash Sehwag

Thanks for the prompt response. I am happy to include the evaluation script following the template. Just to make sure, in the template, robust accuracy refers to numbers calculated with...

Just to follow up on this thread, one suggestion will be to set up a leaderboard for adversarial robustness. Currently, we are missing a strong leaderboard, which is widely recognized,...

I think a dedicated webpage (similar to the spirit of https://paperswithcode.com/sota/image-classification-on-imagenet) will serve well. Maybe paperswithcode.com can host one for adversarial defenses based on your auto-attack.

No worries. I managed to distill the interactive_demo.py to a jupyter notebook without the dependency on bimpy. Its not interactive but easily serve the same purpose with some loops and...

The notebook without bimpy dependency is available at https://github.com/VSehwag/ALAE/blob/master/replicate_results.ipynb. So far, I am quite intrigued by the visualization obtained across a diverse set of images. However, as I mentioned earlier...

In my attempt to reproduce the results, I did a hyperparameter sweep for learning rate, momentum, and weight decay (2-3 values for each). Though the standard setting of (lr=0.1, wd=1e-4,...

Hi, I used a custom celebA conditioned dataset to train this model. This was done by using the following two attributes: 1) gender (male/female) and 2) Smiling/not-smiling. This is the...

Hi @prajwalsingh This is a perfectly natural phenomenon in the first few epochs (and it goes away after 5-10 epochs). I have noticed that the model starts with generating highly...