Stop memorizing: A data-dependent regularization framework for intrinsic pattern learning
This addresses the issue of overfitting and memorization in deep learning for researchers and practitioners, offering a more robust regularization method.
The paper tackles the problem of deep neural networks memorizing random data despite conventional regularization, by proposing a new framework called GRSVNet that enforces and validates geometry consistency during training. The result is that OLE-GRSVNet outperforms regularized DNNs on real data and refuses to memorize random data, indicating it learns intrinsic patterns.
Deep neural networks (DNNs) typically have enough capacity to fit random data by brute force even when conventional data-dependent regularizations focusing on the geometry of the features are imposed. We find out that the reason for this is the inconsistency between the enforced geometry and the standard softmax cross entropy loss. To resolve this, we propose a new framework for data-dependent DNN regularization, the Geometrically-Regularized-Self-Validating neural Networks (GRSVNet). During training, the geometry enforced on one batch of features is simultaneously validated on a separate batch using a validation loss consistent with the geometry. We study a particular case of GRSVNet, the Orthogonal-Low-rank Embedding (OLE)-GRSVNet, which is capable of producing highly discriminative features residing in orthogonal low-rank subspaces. Numerical experiments show that OLE-GRSVNet outperforms DNNs with conventional regularization when trained on real data. More importantly, unlike conventional DNNs, OLE-GRSVNet refuses to memorize random data or random labels, suggesting it only learns intrinsic patterns by reducing the memorizing capacity of the baseline DNN.