2 papers across 2 sessions
Equivariant neural networks can fail to approximate target functions even when they achieve maximal separation power.