Reducing Overfitting in Deep Networks by Decorrelating Representations
This addresses the problem of overfitting for practitioners training deep networks, offering a novel regularizer that is incremental but effective compared to existing methods like Dropout.
The paper tackles overfitting in deep neural networks by proposing a new regularizer called DeCov that minimizes cross-covariance of hidden activations to encourage diverse representations, resulting in reduced overfitting and often improved generalization performance over Dropout across various datasets and architectures.
One major challenge in training Deep Neural Networks is preventing overfitting. Many techniques such as data augmentation and novel regularizers such as Dropout have been proposed to prevent overfitting without requiring a massive amount of training data. In this work, we propose a new regularizer called DeCov which leads to significantly reduced overfitting (as indicated by the difference between train and val performance), and better generalization. Our regularizer encourages diverse or non-redundant representations in Deep Neural Networks by minimizing the cross-covariance of hidden activations. This simple intuition has been explored in a number of past works but surprisingly has never been applied as a regularizer in supervised learning. Experiments across a range of datasets and network architectures show that this loss always reduces overfitting while almost always maintaining or increasing generalization performance and often improving performance over Dropout.