[Comp-neuro] Paper: Forward propagation closed loop learning