About

Portrait of Diego Martinez Taboada

I'm a fourth-year PhD student in the Department of Statistics and Data Science at Carnegie Mellon University, advised by Prof. Aaditya Ramdas.

Previously, I got to learn about mathematics and statistics at the University of Oxford, Sorbonne Université, and Universidade de Santiago de Compostela. I have also done internships on cool machine learning projects at GResearch and CiTIUS.

Google Scholar LinkedIn CV (PDF)

Education

Background

  • 2022 — present
    Carnegie Mellon University PhD in Statistics and Data Science
  • 2021 — 2022
    University of Oxford MSc in Statistical Science
  • 2017 — 2021
    University of Santiago de Compostela BSc in Mathematics

Research

What I work on

I'm broadly interested in pushing statistics (both foundational and methodological) to develop new tools for inference and learning in modern settings, as well as in understanding modern deep learning systems (why and how do deep networks work?).

These interests span high-dimensional concentration (how do random infinite-dimensional objects — gradients, functions, embeddings — concentrate?), concentration for complex data structures (random matrices and operators that need not commute, as arise e.g. in neural network initialization), sequential decision-making (working with streams of data rather than fixed datasets, as in bandits), causal inference (when the causal target depends in complex ways on the data, as with distributional treatment effects), and multiple testing (where evidence across hypotheses is often dependent and the number of hypotheses can be very large).

Preprints

Working papers

  1. Under review

    Bentkus-type asymptotic e-values

    Diego Martinez-Taboada, Ben Chugg, Aaditya Ramdas

    arXiv
  2. Under review

    Intrinsic-dimension empirical Bernstein inequalities for bounded self-adjoint operators

    Diego Martinez-Taboada, Aaditya Ramdas

    arXiv
  3. Under review

    Intrinsic dimension concentration inequalities for self-adjoint operators

    Diego Martinez-Taboada, Aaditya Ramdas

    arXiv

Publications

Papers

  1. Sharp empirical Bernstein bounds for the variance of bounded random variables

    Diego Martinez-Taboada, Aaditya Ramdas

    International Conference on Machine Learning (ICML), 2026

    arXiv

  2. Nonasymptotic heavy-tailed mean estimation in smooth Banach spaces

    Justin Whitehouse, Ben Chugg, Diego Martinez-Taboada, Aaditya Ramdas

    Stochastic Processes and their Applications, 2026

    arXiv Journal

  3. Vector-valued self-normalized concentration inequalities beyond sub-Gaussianity

    Diego Martinez-Taboada, Tomas Gonzalez-Lara, Aaditya Ramdas

    International Conference on Algorithmic Learning Theory (ALT), 2026

    arXiv Proceedings

  4. Empirical Bernstein in smooth Banach spaces

    Diego Martinez-Taboada, Aaditya Ramdas

    Annals of Applied Probability, 2026

    arXiv

  5. Sequential Kernelized Stein Discrepancy

    Diego Martinez-Taboada, Aaditya Ramdas

    International Conference on Artificial Intelligence and Statistics (AISTATS), 2025

    arXiv Proceedings

  6. Counterfactual Density Estimation using Kernel Stein Discrepancies

    Diego Martinez-Taboada, Edward H. Kennedy

    International Conference on Learning Representations (ICLR), 2024

    arXiv Proceedings

  7. An Efficient Doubly-Robust Test for the Kernel Treatment Effect

    Diego Martinez-Taboada, Aaditya Ramdas, Edward H. Kennedy

    Neural Information Processing Systems (NeurIPS), 2023

    arXiv Proceedings

Contact

Get in touch

Email

diegomar [at] andrew [dot] cmu [dot] edu