About Me

I am Harsh, fouth final year PhD student at Department of Computational and Data Science (CDS) at Indian Institute of Science. I am very fortunate to be advised by Prof. Venkatesh Babu. I also closely collaborate with Dr. Varun Jampani and Dr. Sho Takemori. Currently, I am working on the intersection of Machine Learning and Computer Vision. I am supported by Prime Minister’s Research Fellowship in Computer Science. I previously completed an honors degree in Computer Science being jointly advised by Dr. Rajeev Sangal and Dr. A.K. Singh at Department of Computer Science, IIT BHU Varanasi. In the past I have interned with Amazon in Outbound Marketing Automtation team and also have been the lead maintainer of the placement portal site. In my free time I like to read books and play basketball.

News

Feb, 2024 : Our paper “DeiT-LT:Distillation Strikes Back for Vision Transformer Training on Long-Tailed Datasets” got accepted to CVPR'24. Preprint to be out soon.

Feb, 2024 : Invited Talk on “Learning from Limited and Imperfect Data” at IEEE Deep-Tech Seminar.

Feb, 2024 : Serving as reviewer in ECCV'24.

January, 2024 : Our paper on “Selective Mixup Fine-Tuning for Optimizing Non-Decomposable Metrics” got accepted to ICLR'24 for a Spotlight (Top-5%) presentation. Pre-print is available here.

December, 2023 : Serving as reviewer for ICML'24.

November, 2023 : Organizing the Adobe-IISc GenAI Workshop

October, 2023 : Our paper on “Inducing Smoothness Regularization in Federated Learning” got accepted to WACV'24. Pre-print is available here.

October, 2023 : Attended ICCV'23 in Paris. Thanks to Google and Kotak AI Centre for travel grants.

September, 2023 : Serving as reviewer for AAAI'24 and ICLR' 24.

see all news

Research Interests:

I am broadly intereseted in Deep Learning algorithms that train and generalize well on real-world datasets, having presence of class-imbalances and distribution shifts. I use deep learning frameworks like PyTorch, Tensorflow etc. to develop algorithms and techniques from statistics, optimization theory, and information theory to analyze them. I am currently exploring applications in Computer Vision. Below are some of the broad areas I have worked on (further details in Research Statement):

  • Generative Models
  • Long Tail Learning
  • Domain Adaptation of Models
  • Loss Landscape and Optimization

Here is my detailed CV and a research statement RS. Please email me if you are interested in a research collaboration.

Contact me

harsh.rangwani