Portrait of Mahdi Ghaznavi

Mahdi Ghaznavi

Supervisor
Research Topics
AGI (Artificial General Intelligence)
Deep Learning
Foundation Models
Generalization
Optimization
Out-of-Distribution (OOD) Generalization
Representation Learning

Publications

The Geometry of Spectral Gradient Descent: Layerwise Criteria for SignSGD vs SpecSGD
Optimization in deep learning has expanded beyond Euclidean methods to include entrywise sign updates (SignSGD) and spectral sign updates (S… (see more)pecGD/Muon). While both can be viewed as steepest descent under non-Euclidean geometries (