About

Hello!, I am a computer science Ph.D. candidate at the University of Geneva, working under the supervision of François Fleuret. My research focuses on efficient deep learning — model compression, fast surrogate models for simulation in high-energy physics, and geometric deep learning. I am passionate about advancing the state-of-the-art in deep learning and its applications in various domains.

Selected Publications

• 2026
  AA-SVD: Anchored and Adaptive SVD for Large Language Model Compression

  Atul Kumar Sinha, François Fleuret

  arXiv preprint. arXiv:2604.02119, 2026

  Paper Code Bib

  @misc{sinha2026aasvd,
    title        = {AA-SVD: Anchored and Adaptive SVD for Large Language Model Compression},
    author       = {Atul Kumar Sinha and Fran{\c{c}}ois Fleuret},
    year         = {2026},
    eprint       = {2604.02119},
    archivePrefix = {arXiv},
    primaryClass = {cs.LG}
  }

• 2024
  DeepEMD: A Transformer-based Fast Estimation of the Earth Mover's Distance

  Atul Kumar Sinha, François Fleuret

  Topology, Algebra, and Geometry in ML Workshop, ICML, 2023

  International Conference on Pattern Recognition (ICPR). Cham: Springer Switzerland, 2024

  Paper Code Bib

  @inproceedings{sinha2024deepemd,
    title     = {DeepEMD: A Transformer-based Fast Estimation of the Earth Mover's Distance},
    author    = {Atul Kumar Sinha and Fran{\c{c}}ois Fleuret},
    booktitle = {International Conference on Pattern Recognition (ICPR)},
    publisher = {Springer},
    year      = {2024}
  }

• 2023
  SUPA: A Lightweight Diagnostic Simulator for Machine Learning in Particle Physics

  Atul Kumar Sinha, Daniele Paliotta, Bálint Máté, John A. Raine, Tobias Golling, François Fleuret

  NeurIPS Datasets and Benchmarks Track, 2023

  Paper Code Bib

  @inproceedings{sinha2023supa,
    title     = {SUPA: A Lightweight Diagnostic Simulator for Machine Learning in Particle Physics},
    author    = {Atul Kumar Sinha and Daniele Paliotta and B{\'a}lint M{\'a}t{\'e} and John A. Raine and Tobias Golling and Fran{\c{c}}ois Fleuret},
    booktitle = {NeurIPS Datasets and Benchmarks Track},
    year      = {2023}
  }