Obstacles to Nonlinear Learning


What are the existing structures and obstacles that make it difficult to implement nonlinear learning?