Gradient Regularization Improves Accuracy of Discriminative Models
This work addresses generalization issues in discriminative models for computer vision, but it is incremental as it builds on known regularization techniques.
The paper tackles the problem of improving classification accuracy in vision tasks by applying gradient regularization to neural networks, showing consistent gains especially with limited training data.
Regularizing the gradient norm of the output of a neural network with respect to its inputs is a powerful technique, rediscovered several times. This paper presents evidence that gradient regularization can consistently improve classification accuracy on vision tasks, using modern deep neural networks, especially when the amount of training data is small. We introduce our regularizers as members of a broader class of Jacobian-based regularizers. We demonstrate empirically on real and synthetic data that the learning process leads to gradients controlled beyond the training points, and results in solutions that generalize well.