sample-image

About Me

I am a Ph.D. Scholar at the Indian Institute of Technology, Delhi (IIT Delhi), where I have recently presented my synopsis and am currently writing my thesis under the guidance of Prof. Sumeet Agarwal and Prof. Prathosh A.P. My research is focused on developing efficient neural networks for deployment on resource-constrained devices for computer vision and NLP tasks.

During my Ph.D, I have contributed to impactful projects in collaboration with organizations such as Cadence, Policy Bazaar, and Samsung. Before Ph.D. I completed M.Tech from Indian Institute of Technology, Dhanbad working in Audio Signal Processing with machine learning.

Research interests: deep learning, model compression, computer vision, NLP, efficient transformers

My CV

Contact Me: oshin.dutta@ee.iitd.ac.in

If you are interested, lets collaborate ! Drop me a message anywhere through the above links.

Published Work

TVA-prune : Efficient LLM Pruning with Global Token-Dependency Awareness and Hardware-Adapted Inference

Oshin Dutta , Ritvik Gupta, Sumeet Agarwal, in Es-FoMo @ ICML 2024. Paper Code

VTrans: Accelerating Transformer Compression with Variational Information Bottleneck based Pruning

Oshin Dutta , Ritvik Gupta, Sumeet Agarwal, Under Review. Paper Code

Search-time Efficient Device Constraints-Aware Neural Architecture Search

Oshin Dutta , Tanu Kanvar, Sumeet Agarwal, in 10th International Conference on Pattern Recognition and Machine Intelligence (PReMI) 2023. Arxiv Paper

A Variational Information Bottleneck Based Method to Compress Sequential Networks for Human Action Recognition

Oshin Dutta , Ayush Srivastava, Jigyasa Gupta, Sumeet Agarwal, and Prathosh A.P., in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2021. Paper Video

Tempo Octave Correction Using Multiclass Support Vector Machine

Oshin Dutta, in Second International Conference on Inventive Communication and Computational Technologies (ICICCT). IEEE , 2018. Paper

Services

  • Conference Reviewer:
    WACV, Women in Machine Learning (WiML), AISTATS, ICML, IJCAI
  • Teaching Assistant:
    • Cognitive and Intelligent Systems (2023)
    • Introduction to Machine Learning (2022)
    • Machine Intelligence and Learning (2021)
    • Introduction to Electrical Engineering (2021)
    • Signal Processing (2014)

Blog

    Presentation of the paper published in WACV 2021