Normalization techniques are essential for accelerating the training and improving the generalization of deep neural networks (DNNs), and have successfully been used in various applications. This paper reviews and comments on the past, present and future of normalization methods in the context of DNN training. We provide a unified picture of the main motivation behind different approaches from the perspective of optimization, and present a taxonomy for understanding the similarities and differences between them. Specifically, we decompose the pipeline of the most representative normalizing activation methods into three components: the normalization area partitioning, normalization operation and normalization representation recovery. In doing so, we provide insight for designing new normalization technique. Finally, we discuss the current progress in understanding normalization methods, and provide a comprehensive review of the applications of normalization for particular tasks, in which it can effectively solve the key issues.

People with diabetes are at risk of developing an eye disease called diabetic retinopathy (DR). This disease occurs when high blood glucose levels cause damage to blood vessels in the retina. Computer-aided DR diagnosis has become a promising tool for the early detection and severity grading of DR, due to the great success of deep learning. However, most current DR diagnosis systems do not achieve satisfactory performance or interpretability for ophthalmologists, due to the lack of training data with consistent and fine-grained annotations. To address this problem, we construct a large fine-grained annotated DR dataset containing 2,842 images (FGADR). Specifically, this dataset has 1,842 images with pixel-level DR-related lesion annotations, and 1,000 images with image-level labels graded by six board-certified ophthalmologists with intra-rater consistency. The proposed dataset will enable extensive studies on DR diagnosis. Further, we establish three benchmark tasks for evaluation: 1. DR lesion segmentation; 2. DR grading by joint classification and segmentation; 3. Transfer learning for ocular multi-disease identification. Moreover, a novel inductive transfer learning method is introduced for the third task. Extensive experiments using different state-of-the-art methods are conducted on our FGADR dataset, which can serve as baselines for future research. Our dataset will be released in https://github.com/csyizhou/FGADR-2842-Dataset.

Batch Normalization (BN) is capable of accelerating the training of deep models by centering and scaling activations within mini-batches. In this work, we propose Decorrelated Batch Normalization (DBN), which not just centers and scales activations but whitens them. We explore multiple whitening techniques, and find that PCA whitening causes a problem we call stochastic axis swapping, which is detrimental to learning. We show that ZCA whitening does not suffer from this problem, permitting successful learning. DBN retains the desirable qualities of BN and further improves BN's optimization efficiency and generalization ability. We design comprehensive experiments to show that DBN can improve the performance of BN on multilayer perceptrons and convolutional neural networks. Furthermore, we consistently improve the accuracy of residual networks on CIFAR-10, CIFAR-100, and ImageNet.

Batch Normalization (BN) is ubiquitously employed for accelerating neural network training and improving the generalization capability by performing standardization within mini-batches. Decorrelated Batch Normalization (DBN) further boosts the above effectiveness by whitening. However, DBN relies heavily on either a large batch size, or eigen-decomposition that suffers from poor efficiency on GPUs. We propose Iterative Normalization (IterNorm), which employs Newton’s iterations for much more efficient whitening, while simultaneously avoiding the eigen-decomposition. Furthermore, we develop a comprehensive study to show IterNorm has better trade-off between optimization and generalization, with theoretical and experimental support. To this end, we exclusively introduce Stochastic Normalization Disturbance (SND), which measures the inherent stochastic uncertainty of samples when applied to normalization operations. With the support of SND, we provide natural explanations to several phenomena from the perspective of optimization, e.g., why group-wise whitening of DBN generally outperforms full-whitening and why the accuracy of BN degenerates with reduced batch sizes. We demonstrate the consistently improved performance of IterNorm with extensive experiments on CIFAR-10 and ImageNet over BN and DBN.

Noncontrastive self-supervised learning methods offer an effective alternative to contrastive approaches by avoiding the need for negative samples to avoid representation collapse. Noncontrastive learning methods explicitly or implicitly optimize the representation space, yet they often require large representation dimensions, leading to dimensional inefficiency. To provide negative samples, contrastive learning methods often require large batch sizes, thus regarded as sample inefficient, while noncontrastive learning methods require large representation dimensions, thus regarded as dimension inefficient. Although we have some understanding of the noncontrastive learning method, theoretical analysis of such phenomenon still remains largely unexplored. We present a theoretical analysis of the dimensional need for noncontrastive learning. We investigate the transfer between upstream representation learning and downstream tasks’ performance, demonstrating how noncontrastive methods implicitly increase interclass distances within the representation space and how the distance affects the model performance of evaluation performance. We prove that the performance of noncontrastive methods is affected by the output dimension and the number of latent classes, and illustrate why performance degrades significantly when the output dimension is substantially smaller than the number of latent classes. We demonstrate our findings through experiments on image classification experiments, and enrich the verification in audio, graph and text modalities. We also perform empirical evaluation for image models on extensive detection and segmentation tasks beyond classification that show satisfactory correspondence to our theorem.

Orthogonality is widely used for training deep neural networks (DNNs) due to its ability to maintain all singular values of the Jacobian close to 1 and reduce redundancy in representation. This paper proposes a computationally efficient and numerically stable orthogonalization method using Newton's iteration (ONI), to learn a layer-wise orthogonal weight matrix in DNNs. ONI works by iteratively stretching the singular values of a weight matrix towards 1. This property enables it to control the orthogonality of a weight matrix by its number of iterations. We show that our method improves the performance of image classification networks by effectively controlling the orthogonality to provide an optimal tradeoff between optimization benefits and representational capacity reduction. We also show that ONI stabilizes the training of generative adversarial networks (GANs) by maintaining the Lipschitz continuity of a network, similar to spectral normalization (SN), and further outperforms SN by providing controllable orthogonality.