Rayan Harfouche (2022)

Internship project topic: Scalable and Universal Second Order Methods for Convex Optimization

Research Interests

  • Statistical Physics  
  • Optimization
  • Information theory

Biography

I have done a Bachelor in Physics and currently I am a Master student in Applied Physics at EPFL. Physics deals with all kinds of phenomena, but I am particularly interested in how it appears in complex systems, such as a large number of interacting particles or a large data set. These systems, although sometimes governed by simple laws, can result in complicated behaviours that are universal and can be generalized to other systems. This is what statistical physics allows us to study.

Robust NAS under adversarial training: benchmark, theory, and beyond

Y. Wu; F. Liu; C-J. Simon-Gabriel; G. Chrysos; V. Cevher 

2024. 12th International Conference on Learning Representations (ICLR 2024), Vienna, Austria, May 7-11, 2024.

Revisiting Character-level Adversarial Attacks for Language Models

E. Abad Rocamora; Y. Wu; F. Liu; G. Chrysos; V. Cevher 

2024. 41st International Conference on Machine Learning (ICML 2024), Vienna, Austria, July 21-27, 2024.

On the Convergence of Encoder-only Shallow Transformers

Y. Wu; F. Liu; G. Chrysos; V. Cevher 

2023. 37th Annual Conference on Neural Information Processing Systems, New Orleans, USA, December 10-16. 2023.

Extrapolation and Spectral Bias of Neural Nets with Hadamard Product: a Polynomial Net Study

Y. Wu; Z. Zhu; F. Liu; G. Chrysos; V. Cevher 

2022. 36th Conference on Neural Information Processing Systems (NeurIPS), New Orleans, USA, November 28 – December 3, 2022.

Contact