☰

• AI for Statistical Learning Theory (in Lean 4)

• The varphi curve: Generalization under scaling via norm-based capacities

• One-step full gradient suffices for low-rank fine-tuning, provably and efficiently

• Learning with norm-based neural networks: model capacity, function spaces, and computational-statistical gaps

• From kernel methods to neural networks: double descent, function spaces, curse of dimensionality

• Can we avoid robust overfitting in adversarial training? - An approximation viewpoint (regression part)