Kayhan Behdin

alt text 

PhD Candidate,
Operations Research,
Massachusetts Institute of Technology (MIT)
77 Massachusetts Avenue, Cambridge, MA 02139, USA
Email: behdink [@] mit [DOT] edu

About me

I received my B.Sc. in Electrical Engineering from Sharif University of Technology, Tehran, Iran, in 2019. I joined MIT as a PhD student in operation research in 2019. I've been working under the supervision of Prof. Rahul Mazumder since then. I was an AI/ML Engineering Intern at LinkedIn Corporation from June 2022 through August 2022, and from May 2023 through August 2023.

Research

My research interests include

  • Applied Statistics

  • High-dimensional Probability

  • Mathematical Optimization (Convex and Discrete Optimization)

  • Machine Learning

Recent Publications

  1. K. Behdin, A. Acharya, A. Gupta, S. Keerthi and R. Mazumder "QuantEase: Optimization-based Quantization for Language Models – An Efficient and Intuitive Algorithm ", Preprint, 2023. [arxiv]

  2. K. Behdin, W. Chen and R. Mazumder "Sparse Gaussian Graphical Models with Discrete Optimization: Computational and Statistical Perspectives ", Preprint, 2023. [arxiv]

  3. K. Behdin and R. Mazumder "On Statistical Properties of Sharpness-Aware Minimization: Provable Guarantees ", Preprint, 2023. [arxiv]

  4. G. Loewinger, K. Behdin, K. T. Kishida, G. Parmigiani, R. Mazumder "Multi-Task Learning for Sparsity Pattern Heterogeneity: A Discrete Optimization Approach ", Preprint, 2022. [arxiv]

  5. K. Behdin, Q. Song, A. Gupta, D. Durfee, A. Acharya, S. Keerthi, R. Mazumder "Improved Deep Neural Network Generalization Using m-Sharpness-Aware Minimization ", NeurIPS OPT Workshop, 2022. [Paper]

  6. K. Behdin and R. Mazumder " Sparse PCA: A New Scalable Estimator Based On Integer Programming ", Preprint, 2021. [arxiv]

  7. K. Behdin and R. Mazumder " Archetypal Analysis for Sparse Nonnegative Matrix Factorization: Robustness Under Misspecification ", Preprint, 2021. [arxiv]


Google Scholar.
Full list of publications.
CV (Sep 23).