Ohad Shamir

Ohad Shamir is a Professor in the Department of Computer Science and Applied Mathematics at the Weizmann Institute of Science's Faculty of Mathematics and Computer Science. He holds visiting positions at Google Research and Microsoft Research, and completed his PhD at the Hebrew University in 2010. His office is located in Ziskind 254 at the Weizmann campus in Rehovot, Israel.

His educational background includes a PhD from the Hebrew University (2010), with subsequent visiting positions at major tech research labs. He has supervised numerous graduate students, many of whom have become faculty members at prestigious institutions including Weizmann Institute, Hebrew University, and Ben Gurion University.

Shamir's research focuses on theoretical aspects of machine learning, particularly deep learning theory, the intersection of machine learning and optimization, and learning under information and communication constraints. His work bridges mathematical rigor with practical machine learning challenges, examining fundamental questions about neural network expressivity, optimization landscapes, and generalization behavior. He investigates how architectural choices (like depth and width) affect learning capabilities, and explores the theoretical foundations of modern learning algorithms.

His publication record shows consistent high-impact output with recent papers (2023-2025) concentrated in neural network theory, optimization complexity, and statistical learning foundations. Key themes include depth-width tradeoffs, benign overfitting phenomena, nonsmooth optimization challenges, and the theoretical limits of learning with neural networks. His work often provides rigorous mathematical analysis of empirical phenomena observed in deep learning.

Shamir actively contributes to the academic community through teaching advanced courses including Topics in Machine Learning Theory (offered repeatedly since 2013) and Seminar on Deep Learning Theory. He serves as a coordinator for the Weizmann Machine Learning and Statistics Seminar held alternate Wednesdays in Ziskind building. Notably, he will be on sabbatical starting August 2025 and will not host new students or visitors during this period.

His current research group includes PhD students Daniel Barzilai (co-advised with Ronen Basri) and Guy Kornowski, along with MSc students Itamar Shoshani and Amitsour Egosi. His past students include prominent researchers like Gal Vardi (now Weizmann faculty) and Yossi Arjevani (Hebrew University faculty), demonstrating his significant impact on training the next generation of theoretical ML researchers.