Destructive learning is a technique that destroys the initial neural network architecture for the purpose of better learning. One method in this class is structural learning with forgetting .