معرفی
Michael Mahoney is an Adjunct Professor in the Department of Statistics at University of California, Berkeley. He is also affiliated with the International Computer Science Institute (ICSI) as Vice President and Director of the Big Data Group, and the Lawrence Berkeley National Laboratory (LBNL) as Group Lead of the Machine Learning and Analytics Group. His research focuses on Randomized Numerical Linear Algebra (RandNLA), Machine Learning, and Scientific Computing, with applications in genetics, climate science, social networks, and physics-informed models. Mahoney leads initiatives such as the Foundations of Data Analysis (FODA) Institute and develops tools like RandBLAS/RandLAPACK for scalable linear algebra.
Research Interests
Mahoney’s work bridges theoretical and applied domains. Key areas include:
- RandNLA: Leveraging randomness for efficient matrix computations, with implementations in distributed systems.
- Deep Learning Theory: Exploring implicit regularization, heavy-tailed dynamics, and self-organization in neural networks.
- Scientific ML: Combining domain-driven models (e.g., PDEs) with data-driven approaches for physical systems like fluid dynamics and seismology.
Teaching & Mentorship
He teaches advanced courses on Linear Algebra for Data Science and Randomized Linear Algebra, emphasizing foundational methods relevant to modern ML. Mahoney has advised numerous PhD and Master’s students across disciplines, fostering innovation in scalable algorithms and their applications.
Grants & Collaborations
Mahoney’s funding includes NSF, DOE, and DARPA grants for projects like Scientific Foundation Models and Heavy-Tailed ML Dynamics. His collaborations span academia (e.g., MIT, Stanford) and industry (Amazon, NVIDIA), focusing on computational efficiency and real-world impact.
