• 2024 - 2029 | PhD in Computer Science | Princeton University

  • Studying Theoretical Machine Learning under the supervision of Prof. Elad Hazan.

• 2023 - 2024 | M.Sc. in Electrical and Computer Engineering | Technion

  • Studied theory for Deep Learning under the supervision of Prof. Daniel Soudry.

  • Focused on generalization bounds for neural networks, in the setting of random interpolation.

• 2020 - 2024 | B.Sc. in Computer Science and Physics | Technion

  • Studied Deep Learning under the supervision of Prof. Michal Irani.

  • Focused on memorization in neural networks, using training data reconstruction.

• How Uniform Random Weights Induce Non-uniform Bias: Typical Interpolating Neural Networks Generalize with Narrow Teachers

Buzaglo*, Harel*,  Shpigel-Nacson*,  Brutzkus, Srebro and Soudry. ICML 2024

• recorded talk (1 hour), starts at 05:45 

• Deconstructing Data Reconstruction: Multiclass, Weight Decay and General Losses

Buzaglo*, Haim*,Yehudai, Vardi, Oz, Nikankin and Irani. NeurIPS 2023

• • 5 min video | GitHub project | preliminary version in TrustML-(un)Limited Workshop @ ICLR 2023

• Continual Learning in Linear Classification on Separable Data

Evron, Moroshko, Buzaglo, Khriesh, Marjieh, Srebro and Soudry. ICML 2023

• 15 min video