mohammad hossein amani

(but people usually call me masani)

prof_pic.jpg

Ph.D. student in computer science

EPFL, Switzerland

Hello friend,

Since September 2022 I am a Ph.D. student at EPFL, where I study symbolic representation learning, reinforcement learning, and sequence-to-sequence modeling. Here’s my CV!

Before starting my Ph.D., I spent some time at SPOC, IST Austria, and CUHK where I studied more mathematically themed problems in machine learning.

Outside of work, I like bicycles, books or good writing of almost any kind, peanut butter, genuine conversations, indie songs, dogs, running, steadfastness, climbing, the earthy smell after rain, wholesome vegetrian food, embarking on adventures, and coffee.

I sometimes take photos from things that I find interesting. You can check them out on instagram or unsplash.

selected publications

  1. Symbolic autoencoding for self-supervised sequence learning
    Mohammad Hossein Amani, Nicolas Mario Baldwin, Amin Mansouri, Martin Josifoski, Maxime Peyrard, and Robert West
    under peer review, 2024
    ICML’24 Workshop Code Poster
  2. Sharp asymptotics on the compression of two-layer neural networks
    Mohammad Hossein Amani, Simone Bombari, Marco Mondelli, Rattana Pukdee, and Stefano Rini
    Proc. of the IEEE Information Theory Workshop (ITW), 2022
    ITW’22
  3. Memorization and optimization in deep neural networks with minimum over-parameterization
    Simone Bombari, Mohammad Hossein Amani, and Marco Mondelli
    Proc. of the Conference on Neural Information Processing Systems (NeurIPS), 2022
    Neurips’22