Behrooz Tahmasebi

Hello! I'm a Ph.D. student in EECS at MIT and a student member of CSAIL. My advisor is Prof. Stefanie Jegelka. Previously, I studied at Sharif University of Technology.

I'm interested in learning with graphs/manifolds/invariances, deep learning (geometry, optimization), and Large Language Models (LLMs) foundations.

CV: click here

Email: bzt at mit dot edu

[Google scholar]

Manuscripts under Review:

Note: authors denoted with * contributed equally.

A Robust Kernel Statistical Test of Invariance: Detecting Subtle Asymmetries [preprint]
    Ashkan Soleymani*, Behrooz Tahmasebi*, Stefanie Jegelka, Patrick Jaillet
    preprint (submitted), 2024

Learning with Exact Invariances in Polynomial Time [preprint]
    Ashkan Soleymani*, Behrooz Tahmasebi*, Stefanie Jegelka, Patrick Jaillet
    preprint (submitted), 2024

Generalization Bounds for Canonicalization: A Comparative Study with Group Averaging [preprint]
    Behrooz Tahmasebi, Stefanie Jegelka
    preprint (submitted), 2024

Regularity in Canonicalized Models: A Theoretical Perspective [preprint]
    Behrooz Tahmasebi, Stefanie Jegelka
    preprint (submitted), 2024

Published Papers:

Coded Computing for Resilient Distributed Computing: A Learning-Theoretic Framework [arXiv]
    Parsa Moradi, Behrooz Tahmasebi, Mohammad Ali Maddah-Ali
    NeurIPS 2024

A Universal Class of Sharpness-Aware Minimization Algorithms [arXiv]
    Behrooz Tahmasebi, Ashkan Soleymani, Dara Bahri, Stefanie Jegelka, Patrick Jaillet
    ICML 2024
    Best Paper Award, workshop on High-dimensional Learning Dynamics (HiLD) at ICML 2024

Sample Complexity Bounds for Estimating Probability Divergences under Invariances [arXiv]
    Behrooz Tahmasebi, Stefanie Jegelka
    ICML 2024

The Exact Sample Complexity Gain from Invariances for Kernel Regression [arXiv]
    Behrooz Tahmasebi, Stefanie Jegelka
    NeurIPS 2023, Spotlight (top 3.6% of submissions)

The Power of Recursion in Graph Neural Networks for Counting Substructures [Conference Link]
    Behrooz Tahmasebi, Derek Lim, Stefanie Jegelka
    AISTATS 2023, Oral (top 1.9% of submissions)

The Capacity of Associated Subsequence Retrieval [Journal Link]
    Behrooz Tahmasebi, Mohammad Ali Maddah-Ali, and Seyed Abolfazl Motahari
    IEEE Transactions on Information Theory, 2021

Private Function Computation
    Behrooz Tahmasebi, Mohammad Ali Maddah-Ali
    IEEE International Symposium on Information Theory (ISIT), 2020

Private Sequential Function Computation [arXiv]
    Behrooz Tahmasebi, Mohammad Ali Maddah-Ali
    IEEE International Symposium on Information Theory (ISIT), 2019

Information Theory of Mixed Population Genome-Wide Association Studies
    Behrooz Tahmasebi, Mohammad Ali Maddah-Ali, Seyed Abolfazl Motahari
    IEEE Information Theory Workshop (ITW), 2018

Genome-Wide Association Studies: Information Theoretic Limits of Reliable Learning
    Behrooz Tahmasebi, Mohammad Ali Maddah-Ali, Seyed Abolfazl Motahari
    IEEE International Symposium on Information Theory (ISIT), 2018

Optimum Transmission Delay for Function Computation in NFV-Based Networks: The Role of Network Coding and Redundant Computing
    Behrooz Tahmasebi, Mohammad Ali Maddah-Ali, Saeedeh Parsaeefard, Babak Khalaj,
    IEEE Journal on Selected Areas in Communications (JSAC), 2018

On the Identifiability of Finite Mixtures of Finite Product Measures
    Behrooz Tahmasebi, Seyed Abolfazl Motahari, Mohammad Ali Maddah-Ali
    Manuscript, 2018

[Accessibility]