publications

2026

  1. TopoFisher: Learning Topological Summary Statistics by Maximizing Fisher Information
    Matteo Biagetti, Mathieu Carrière, Francesco Conti, Enrico Maria Ferrari, Sven Heydenreich, and Karthik Viswanathan
    Advances in Neural Information Processing Systems (NeurIPS), 2026
    arXiv Code
  2. A Transformer Architecture Alteration to Incentivise Externalised Reasoning
    Elizabeth Pavlova*, Mariia Koroliuk*, Karthik Viswanathan*, Cameron Tice, Edward Young, and Puria Radmard
    Under review, 2026
    arXiv Code
  3. The Intrinsic Dimension of Prompts in Internal Representations of Large Language Models
    Karthik Viswanathan, Yuri Gardinazzi, Giada Panerai, Alberto Cazzaniga, and Matteo Biagetti
    Transactions on Machine Learning Research, 2026
    arXiv PDF

2025

  1. Probing Geometry of Next Token Prediction Using Cumulant Expansion of the Softmax Entropy
    Karthik Viswanathan, and Sang Eon Park
    In High-dimensional Learning Dynamics, 2025
    PDF
  2. Persistent Topological Features in Large Language Models
    Yuri Gardinazzi*, Karthik Viswanathan*, Giada Panerai, Alessio Ansuini, Alberto Cazzaniga, and Matteo Biagetti
    Proceedings of the 42nd International Conference on Machine Learning, 2025
    arXiv

2024

  1. Cosmology with persistent homology: a Fisher forecast
    Jacky H. T. Yip, Matteo Biagetti, Alex Cole, Karthik Viswanathan, and Gary Shiu
    Journal of Cosmology and Astroparticle Physics, 2024
    DOI arXiv
  • In machine learning / artificial intelligence papers, an asterisk (*) denotes equal contribution. In theoretical high-energy physics, the convention is to order the author list alphabetically.
  • The Journal of Cosmology and Astroparticle Physics (JCAP) is a leading journal in the field (see, e.g., Google Scholar’s Astronomy & Astrophysics top publications).