Scholar
Emanuele Francazi
Google Scholar ID: hf5I62kAAAAJ
PhD student, École Polytechnique Fédérale de Lausanne
Citations & Impact
Citations
44
H-index
3
i10-index
1
Publications
6
Co-authors
2
list available
Publications
Resume (English only)
Academic Achievements
- Published papers: 'Initial Guessing Bias: How Untrained Networks Favor Some Classes' (ICML 2024); 'Where You Place the Norm Matters: From Prejudiced to Neutral Initializations' (arXiv); 'When the Left Foot Leads to the Right Path: Bridging Initial Prejudice and Trainability' (arXiv); 'A Theoretical Analysis of the Learning Dynamics under Class Imbalance' (ICML 2023).
Research Experience
- Collaborated with Marco Baity-Jesi, Florent Krzakala, Aurelien Lucchi, Giulio Biroli, Marc Mézard, and Jean-Philippe Bouchaud to study biases in neural networks, learning processes, and methods to enhance the efficiency of generative models.
Education
- PhD candidate at EPFL; previously studied theoretical physics at La Sapienza University of Rome, working with Giorgio Parisi and Federico Ricci Tersenghi on spin glasses and critical phenomena in graphs.
Background
- Research interests include predictive bias and learning dynamics in neural networks, aiming to improve the reliability and efficiency of deep learning systems through theoretically grounded approaches. Recent research has investigated how biases emerge in neural networks, how they affect learning, and how principled design choices can help control them.
Miscellany
- Actively seeking an internship to leverage and expand skills in innovative research environments.
