Compositional Program Induction
Tackling discrete compositional program induction with latent structures
Program InductionLatticeCompositional Reasoning
This project explores how geometric deep learning principles can be applied to program induction tasks. The key insight is that programs have inherent symmetries that neural networks can exploit.