CS PhD Student, CU Boulder
Email: nidhin.harilal@colorado.edu
  CV   |     LinkedIn   |     Scholar   |     GitHub

About
I am a PhD student in computer science at the University of Colorado Boulder, where I am advised by Dr. Claire Monteleoni. My research interests are in developing robust and interpretable AI, with a focus on self-supervised models. I have worked with generative AI models for synthetic pre-training and data attribution methods for transparency. I am also collaborating on AI advances for climate science initiatives, particularly in enhancing the resolution and analysis of spatiotemporal climate data.

I am a recipient of Bell Research Fellowship (2024) and, Awtar Teji Singh Fellowship (2022). My reseach on AI for climate is supported by iHARP. I recieved my Bachelor’s degree with honours in Computer Science & Engineering in 2021 from IIT Gandhinagar, where I have been mentored by Auroop Ganguly, Udit Bhatia on AI interpretability and Shanmuganathan Raman on video generation.
Updates
  • [Dec 2024]   InfluenceSSL presented at NeurIPS 2024 - Self-supervised learning & Theory.
  • [Nov 2024]   Recieved Bell Foundation Research Fellowship.
  • [Oct 2024]   MixDiff accepted at WACV 2025. Check it out here!
  • [Aug 2024]  Invited for a talk at Google DeepMind, Paris.
  • [Jun 2024]   PEFT ViT presented at CVPR 2024 - efficient Large Vision models.
  • [May 2024]   Joining INRIA, Paris as a Research Intern.
My Research
2024
MixDiff: Mixing Natural and Synthetic Images for Robust Self-Supervised Representations.
Reza Akbarian* Nidhin Harilal*, Claire Monteleoni, and Maziar Raissi.
Accepted at IEEE/CVF WACV 2025.    PDF     Code    Web
Influence Estimation in Self-Supervised Learning.
Nidhin Harilal*, Reza Akbarian*, Amit Kiran Rege*, Claire Monteleoni, and Maziar Raissi.
In NeurIPS 2024 - Self-Supervised Learning-Theory and Practice.   PDF     Code
Parameter Efficient Fine-tuning of Self-supervised ViTs without Catastrophic Forgetting.
Reza Akbarian*, Nidhin Harilal*, Claire Monteleoni, and Maziar Raissi.
In CVPR 2024 - Efficient Large Vision Models (eLVM).   PDF      Code
2023
STint: Self-supervised Temporal Interpolation for Geospatial Data.
Nidhin Harilal, B. M Hodge, Claire Monteleoni, and Aneesh Subramanian.
In arXiv preprint arXiv:2309.00059 (2023).   PDF
2022
EnhancedSD: Predicting Solar Power Reanalysis from Climate Projections via Image Super-Resolution
Nidhin Harilal, B. M Hodge, Claire Monteleoni, and Aneesh Subramanian.
In NeurIPS 2022 - Tackling Climate Change with Machine Learning.   PDF    Code
Effectiveness of the Recent Advances in Capsule Networks.
Nidhin Harilal, and Rohan Patil.
In arXiv preprint arXiv:2210.05834 (2022)   PDF
Image caption generator using siamese graph convolutional networks and LSTM.
Athul Kumar, Aarchi Agrawal, KS Ashin Shanly, Sudip Das, and Nidhin Harilal.
In ACM IKDD CODS and COMAD 2022.   PDF
2021
HDRVideo-GAN: deep generative HDR video reconstruction.
Anand, Mrinal, Nidhin Harilal, Chandan Kumar, and Shanmuganathan Raman.
In ICVGIP 2021.   PDF
Bayesian Deep Learning Hyperparameter Search for Robust Function Mapping to Polynomials with Noise.
Nidhin Harilal, Udit Bhatia, and Auroop R. Ganguly.
In arXiv preprint arXiv:2106.12532 2021.   PDF
Augmented convolutional LSTMs for generation of high-resolution climate change projections.
Nidhin Harilal, Mayank Singh, and Udit Bhatia.
In IEEE Access 9 2021: 25208-25218.   PDF
CARO: an empathetic health conversational chatbot for people with major depression.
Nidhin Harilal, Rushil Shah, Saumitra Sharma, and Vedanta Bhutani.
In ACM IKDD CoDS and 25th COMAD 2020.   PDF