Aryan Mokhtari

Assistant Professor, Dept. of Electrical & Computer Engineering (ECE), UT Austin
Jack Kilby/Texas Instruments Endowed Faculty Fellowship in Computer Engineering
Affiliated with:
- - NSF AI Institute for Foundations of Machine Learning (IFML)
  - NSF AI Institute for Future Edge Networks and Distributed Intelligence (AI-EDGE)
Google Scholar / Twitter

News

July ’24: Qiujiang (joining Goldman Sachs), Isidoros (joining Meta), Liam (joining Snap Research), and Devyani (joining Goldman Sachs) have graduated! Wishing them all the best as they embark on their new journeys!
May ’24: Paper at ICML 2024 accepted for Oral Presentation:
- Provable Multi-Task Representation Learning by Two-Layer ReLU Neural Networks
Apr. ’24: Google Research Scholar Award. (Announcement)
Apr. ’24: New results on Global Convergence Analysis of BFGS:
- Inexact line search: “Non-asymptotic Global Convergence Analysis of BFGS with the Armijo-Wolfe Line Search” [pdf]
- Exact line search: “Non-asymptotic Global Convergence Rates of BFGS with Exact Line Search” [pdf]
Feb. ’24: New Talk: “In-Context Learning with Transformers: Softmax Attention Adapts to Function Lipschitzness” [Sildes]
- Based on the following paper: [pdf]
Jan. ’24: NSF CAREER Award.
Jan. ’24: Paper at AISTATS 2024:
- Krylov Cubic Regularized Newton: A Subspace Second-Order Method with Dimension-Free Convergence Rate
Oct. ’23: New Talk: “Online Learning Guided Quasi-Newton Methods: Improved Global Non-asymptotic Guarantees” [Slides]
- Based the following COLT 2023 and NeurIPS 2023 papers.
Sep. ’23: Papers at NeurIPS 2023:
May ’23: Papers at COLT 2023:
- Online Learning Guided Curvature Approximation: A Quasi-Newton Method with Global Non-Asymptotic Superlinear Convergence
- InfoNCE Loss Provably Learns Cluster-Preserving Representations

Support

My research has been supported by an NSF CAREER Award CCF-2338846, NSF Grants CCF-2007668 and ECCS-2127697, an ARO Early Career Program (ECP) Award, a Google Research Scholar Award, the NSF AI Institute for Foundations of Machine Learning (IFML), the NSF AI Institute for Future Edge Networks and Distributed Intelligence (AI-EDGE), and the Machine Learning Laboratory at UT Austin.