Principal Researcher, Fondazione Bruno Kessler
2 papers at NeurIPS 2025
Equivariant neural networks can fail to approximate target functions even when they achieve maximal separation power.