Siba Smarak Panigrahi
 Email: siba.panigrahi@epfl.ch

CV | Google Scholar | Semantic Scholar | Github | Twitter | LinkedIn | Instagram

More about me: Three Minute Reads - Classic Short Stories | Fun/Favorites

I am a Ph.D. at École Polytechnique Fédérale de Lausanne (EPFL) working with Prof. Maria Brbic in the Brbic Lab. My research interests span geometric deep learning, generative models, and computer vision.

Previously, I was a visiting researcher at ServiceNow Research in the Multimodal Foundation Models team, building foundation model for structured document understanding. I was also a Research Intern at PROSE at Microsoft, working on designing algorithms and evaluation set-ups for email classification in real-world (online) settings. I was fortunate to work on topological data analysis at Adobe Research, India and on explainability in pre-trained language models at INK-Lab, University of Southern California under Prof. Xiang Ren.

I completed my M.Sc. in Computer Science from McGill University in a thesis-based program under the supervision of Prof. Siamak Ravanbakhsh. I was also affiliated with Mila during this time. I earned my B. Tech. in Computer Science and Engineering from Indian Institute of Technology Kharagpur. I was part of the CVIR Lab under the supervision of Prof. Abir Das and Dr. Rameswar Panda where I worked on contextual bias and multimodal learning problems.

Other: Advice on Grad Applications | Montreal, Canada | Banff, Canada |
ML Reproducibility Challenge (MLRC) | Summer Schools

Other Other: Lore Podcast Summaries (help needed!)

  Recent News
     [Oct 24] One paper accepted in NeurIPS 2024 Workshop on RBFM.
     [Oct 24] One paper accepted in NeurIPS 2024 AI4Mat workshop with a spotlight talk.
     [Sep 24] Received EDIC Fellowship offer to start my Ph.D. at EPFL!
     [Jul 24] One blogpost accepted in GRaM workshop, ICML 2024! Blogpost here.
     [May 24] Two papers accepted in EquiVision workshop, CVPR 2024 with a spotlight talk! Slides here.
     [Apr 24] Gave a talk on Equivariant Adaptation of Large Pretrained Models at Google Research India. Slides here.
     [Mar 24] Received GREAT award from McGill University to attend ICLR 2024.
     [Mar 24] Released open-source Python package EquiAdapt! Feedback is welcome.
     [Jan 24] One paper accepted in ICLR 2024!
     [Oct 23] Organizing ML Reproducibility Challenge 2023!
     [Oct 23] Sponsored CA$ 6000 Google Cloud credits for CampusPulse Ideathon with KDAG, IIT Kharagpur!
  Publications
ImprovedEquiPretrainingCVPR2024

Improved Canonicalization for Model Agnostic Equivariance
Siba Smarak Panigrahi, Arnab Kumar Mondal
Equivariant Vision (EquiVision) workshop (CVPR) 2024

KoopmanRLarXiv2023

Efficient Dynamics Modeling in Interactive Environments with Koopman Theory
Arnab Kumar Mondal, Siba Smarak Panigrahi, Sai Rajeswar, Kaleem Siddiqi, Siamak Ravanbakhsh
International Conference on Learning Representations (ICLR) 2024; European Workshop on Reinforcement Learning (EWRL) 2023

pdf
EquiPretrainingNeurips2023

Equivariant Adaptation of Large Pretrained Models
Arnab Kumar Mondal*, Siba Smarak Panigrahi*, Sékou-Oumar Kaba, Sai Rajeswar, Siamak Ravanbakhsh
Conference on Neural Information Processing Systems (NeurIPS) 2023

VavRC2021

[Re]: Value Alignment Verification
Siba Smarak Panigrahi*, Sohan Patnaik*
The ML Reproducibility Challenge (MLRC), 2021;
NeurIPS Journal Showcase Track, 2022; NeurIPS Spotlight, 2022

pdf / code
ArgminingEMNLP2021

Leveraging Pretrained Language Models for Key Point Matching
Manav Nitin Kapadnis*, Sohan Patnaik*, Siba Smarak Panigrahi*, Varun Madhavan*, and Abhilash Nandy
8th workshop on ArgumentMining at Empirical Methods in Natural Language Processing (EMNLP), 2021

pdf / code
EmotionIACC2020

Multi-class Emotion Classification Using EEG Signals
Divya Acharya, Riddhi Jain, Siba Smarak Panigrahi, Rahul Sahni, Siddhi Jain, Sanika Prashant Deshmukh, and Arpit Bhardwaj
10th International Advance Computing Conference (IACC), 2020

pdf / code
  Packages
EquiadaptGitHub2024

EquiAdapt
Library to make any existing neural network architecture equivariant.


Template: this