{"url":"/method/weight-standardization","slug":"weight-standardization","name":"Weight Standardization","full_name":"Weight Standardization","full_name_withheld":false,"description_markdown":"**Weight Standardization** is a normalization technique that smooths the loss landscape by standardizing the weights in convolutional layers. Different from the previous normalization methods that focus on *activations*, WS considers the smoothing effects of *weights* more than just length-direction decoupling. Theoretically, WS reduces the Lipschitz constants of the loss and the gradients.\r\nHence, WS smooths the loss landscape and improves training.\r\n\r\nIn Weight Standardization, instead of directly optimizing the loss $\\mathcal{L}$ on the original weights $\\hat{W}$, we reparameterize the weights $\\hat{W}$ as a function of $W$, i.e. $\\hat{W}=\\text{WS}(W)$, and optimize the loss $\\mathcal{L}$ on $W$ by [SGD](https://paperswithcode.com/method/sgd):\r\n\r\n$$\r\n    \\hat{W} = \\Big[ \\hat{W}\\_{i,j}~\\big|~ \\hat{W}\\_{i,j} = \\dfrac{W\\_{i,j} - \\mu\\_{W\\_{i,\\cdot}}}{\\sigma\\_{W\\_{i,\\cdot}+\\epsilon}}\\Big]\r\n$$\r\n\r\n$$\r\n    y = \\hat{W}*x\r\n$$\r\n\r\nwhere\r\n\r\n$$\r\n    \\mu_{W\\_{i,\\cdot}} = \\dfrac{1}{I}\\sum\\_{j=1}^{I}W\\_{i, j},~~\\sigma\\_{W\\_{i,\\cdot}}=\\sqrt{\\dfrac{1}{I}\\sum\\_{i=1}^I(W\\_{i,j} - \\mu\\_{W\\_{i,\\cdot}})^2}\r\n$$\r\n\r\nSimilar to [Batch Normalization](https://paperswithcode.com/method/batch-normalization), WS controls the first and second moments of the weights of each output channel individually in convolutional layers. Note that many initialization methods also initialize the weights in some similar ways. Different from those methods, WS standardizes the weights in a differentiable way which aims to normalize gradients during back-propagation. Note that we do not have any affine transformation on $\\hat{W}$. This is because we assume that normalization layers such as BN or [GN](https://paperswithcode.com/method/group-normalization) will normalize this convolutional layer again.","description_state":"present","introduced_year":null,"introduced_by":{"title":"Micro-Batch Training with Batch-Channel Normalization and Weight Standardization","paper":"/paper/weight-standardization","first_author":"Siyuan Qiao","n_authors":5,"url_abs":null,"archive_paper_url":"https://paperswithcode.com/paper/weight-standardization"},"source":{"url":"https://arxiv.org/abs/1903.10520v2","title":"Micro-Batch Training with Batch-Channel Normalization and Weight Standardization","url_on_a_paper_host":true},"code_snippet_url":null,"code_snippet_url_on_a_code_host":false,"categories":[{"area":"General","area_id":"general","collection":"Normalization","url":"/methods/category/normalization","pwc_aliases":[]}],"n_papers_tagged":12,"archive_num_papers":12,"papers_newest_first":[{"paper":null,"title":"Addressing Data Heterogeneity in Federated Learning with Adaptive Normalization-Free Feature Recalibration","date":"2024-10-02","arxiv_id":"2410.02006","n_code_links":0,"syntology":null},{"paper":null,"title":"Adaptive Gradient Regularization: A Faster and Generalizable Optimization Technique for Deep Neural Networks","date":"2024-07-24","arxiv_id":"2407.16944","n_code_links":0,"syntology":null},{"paper":"/paper/exploring-loss-functions-for-time-based","title":"Exploring Loss Functions for Time-based Training Strategy in Spiking Neural Networks","date":"2023-09-21","arxiv_id":null,"n_code_links":1,"syntology":null},{"paper":"/paper/rotational-optimizers-simple-robust-dnn","title":"Rotational Equilibrium: How Weight Decay Balances Learning Across Neural Networks","date":"2023-05-26","arxiv_id":"2305.17212","n_code_links":2,"syntology":{"ran":0,"of":6,"unverified":6,"pointer_only":0}},{"paper":null,"title":"Domain Adaptation and Active Learning for Fine-Grained Recognition in the Field of Biodiversity","date":"2021-10-22","arxiv_id":"2110.11778","n_code_links":0,"syntology":null},{"paper":"/paper/effect-of-large-scale-pre-training-on-full","title":"Effect of Pre-Training Scale on Intra- and Inter-Domain Full and Few-Shot Transfer Learning for Natural and Medical X-Ray Chest Images","date":"2021-05-31","arxiv_id":"2106.00116","n_code_links":1,"syntology":null},{"paper":"/paper/recu-reviving-the-dead-weights-in-binary","title":"ReCU: Reviving the Dead Weights in Binary Neural Networks","date":"2021-03-23","arxiv_id":"2103.12369","n_code_links":3,"syntology":{"ran":2,"of":3,"unverified":1,"pointer_only":3}},{"paper":"/paper/characterizing-signal-propagation-to-close-1","title":"Characterizing signal propagation to close the performance gap in unnormalized ResNets","date":"2021-01-21","arxiv_id":"2101.08692","n_code_links":4,"syntology":{"ran":1,"of":1,"unverified":0,"pointer_only":0}},{"paper":"/paper/monocular-depth-estimation-using-laplacian","title":"Monocular Depth Estimation Using Laplacian Pyramid-Based Depth Residuals","date":"2021-01-08","arxiv_id":null,"n_code_links":1,"syntology":null},{"paper":"/paper/gradient-centralization-a-new-optimization","title":"Gradient Centralization: A New Optimization Technique for Deep Neural Networks","date":"2020-04-03","arxiv_id":"2004.01461","n_code_links":8,"syntology":{"ran":2,"of":29,"unverified":27,"pointer_only":1}},{"paper":"/paper/large-scale-learning-of-general-visual","title":"Big Transfer (BiT): General Visual Representation Learning","date":"2019-12-24","arxiv_id":"1912.11370","n_code_links":9,"syntology":{"ran":3,"of":10,"unverified":7,"pointer_only":0}},{"paper":"/paper/weight-standardization","title":"Micro-Batch Training with Batch-Channel Normalization and Weight Standardization","date":"2019-03-25","arxiv_id":"1903.10520","n_code_links":7,"syntology":null}],"papers_shown":12,"tasks":[{"task":"/task/image-classification","name":"Image Classification","papers":5},{"task":"/task/image-classification","name":"image-classification","papers":3},{"task":"/task/few-shot-learning","name":"Few-Shot Learning","papers":2},{"task":"/task/fine-grained-image-classification","name":"Fine-Grained Image Classification","papers":2},{"task":"/task/transfer-learning","name":"Transfer Learning","papers":2},{"task":"/task/active-learning","name":"Active Learning","papers":1},{"task":"/task/binarization","name":"Binarization","papers":1},{"task":"/task/decoder","name":"Decoder","papers":1},{"task":"/task/depth-estimation","name":"Depth Estimation","papers":1},{"task":"/task/domain-adaptation","name":"Domain Adaptation","papers":1},{"task":"/task/federated-learning","name":"Federated Learning","papers":1},{"task":null,"name":"GPU","papers":1},{"task":"/task/classification","name":"General Classification","papers":1},{"task":"/task/image-generation","name":"Image Generation","papers":1},{"task":"/task/instance-segmentation","name":"Instance Segmentation","papers":1},{"task":"/task/l2-regularization","name":"L2 Regularization","papers":1},{"task":"/task/language-modeling","name":"Language Modeling","papers":1},{"task":"/task/language-modelling","name":"Language Modelling","papers":1},{"task":"/task/monocular-depth-estimation","name":"Monocular Depth Estimation","papers":1},{"task":"/task/object-detection","name":"Object Detection","papers":1}],"tasks_shown":20,"n_tasks":30,"usage_by_year":[{"year":"2019","papers":2},{"year":"2020","papers":1},{"year":"2021","papers":5},{"year":"2023","papers":2},{"year":"2024","papers":2}],"row_source":"methods_table","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/weight-standardization"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}