{"url":"/method/simclr","slug":"simclr","name":"SimCLR","full_name":"SimCLR","full_name_withheld":false,"description_markdown":"**SimCLR** is a framework for contrastive learning of visual representations. It learns representations by maximizing agreement between differently augmented views of the same data example via a contrastive loss in the latent space. It consists of:\r\n\r\n- A stochastic data augmentation module that transforms any given data example randomly resulting in two correlated views of the same example, denoted $\\mathbf{\\tilde{x}\\_{i}}$ and $\\mathbf{\\tilde{x}\\_{j}}$, which is considered a positive pair. SimCLR sequentially applies three simple augmentations: random cropping followed by resize back to the original size, random color distortions, and [random Gaussian blur](https://paperswithcode.com/method/random-gaussian-blur). The authors find random crop and color distortion is crucial to achieve good performance.\r\n\r\n- A neural network base encoder $f\\left(·\\right)$ that extracts representation vectors from augmented data examples. The framework allows various choices of the network architecture without any constraints. The authors opt for simplicity and adopt [ResNet](https://paperswithcode.com/method/resnet) to obtain $h\\_{i} = f\\left(\\mathbf{\\tilde{x}}\\_{i}\\right) = \\text{ResNet}\\left(\\mathbf{\\tilde{x}}\\_{i}\\right)$ where $h\\_{i} \\in \\mathbb{R}^{d}$ is the output after the [average pooling](https://paperswithcode.com/method/average-pooling) layer.\r\n\r\n- A small neural network projection head $g\\left(·\\right)$ that maps representations to the space where contrastive loss is applied. Authors use a MLP with one hidden layer to obtain $z\\_{i} = g\\left(h\\_{i}\\right) = W^{(2)}\\sigma\\left(W^{(1)}h\\_{i}\\right)$ where $\\sigma$ is a [ReLU](https://paperswithcode.com/method/relu) nonlinearity. The authors find it beneficial to define the contrastive loss on $z\\_{i}$’s rather than $h\\_{i}$’s.\r\n\r\n- A contrastive loss function defined for a contrastive prediction task. Given a set {$\\mathbf{\\tilde{x}}\\_{k}$} including a positive pair of examples $\\mathbf{\\tilde{x}}\\_{i}$ and $\\mathbf{\\tilde{x}\\_{j}}$ , the contrastive prediction task aims to identify $\\mathbf{\\tilde{x}}\\_{j}$ in {$\\mathbf{\\tilde{x}}\\_{k}$}$\\_{k\\neq{i}}$ for a given $\\mathbf{\\tilde{x}}\\_{i}$.\r\n\r\nA minibatch of $N$ examples is randomly sampled and the contrastive prediction task is defined on pairs of augmented examples derived from the minibatch, resulting in $2N$ data points. Negative examples are not sampled explicitly. Instead, given a positive pair, the other $2(N − 1)$ augmented examples within a minibatch are treated as negative examples. A [NT-Xent](https://paperswithcode.com/method/nt-xent) (the normalized\r\ntemperature-scaled cross entropy loss) loss function is used (see components).","description_state":"present","introduced_year":null,"introduced_by":{"title":null,"paper":null,"first_author":null,"n_authors":0,"url_abs":null,"archive_paper_url":null},"source":{"url":"https://arxiv.org/abs/2002.05709v3","title":"A Simple Framework for Contrastive Learning of Visual Representations","url_on_a_paper_host":true},"code_snippet_url":"https://github.com/google-research/simclr","code_snippet_url_on_a_code_host":true,"categories":[{"area":"General","area_id":"general","collection":"Self-Supervised Learning","url":"/methods/category/self-supervised-learning","pwc_aliases":[]}],"n_papers_tagged":240,"archive_num_papers":null,"papers_newest_first":[{"paper":null,"title":"Probabilistic Variational Contrastive Learning","date":"2025-06-11","arxiv_id":"2506.10159","n_code_links":0,"syntology":null},{"paper":"/paper/2506-10031","title":"scSSL-Bench: Benchmarking Self-Supervised Learning for Single-Cell Data","date":"2025-06-10","arxiv_id":"2506.10031","n_code_links":1,"syntology":null},{"paper":"/paper/circumventing-backdoor-space-via-weight","title":"Circumventing Backdoor Space via Weight Symmetry","date":"2025-06-09","arxiv_id":"2506.07467","n_code_links":1,"syntology":null},{"paper":"/paper/ssps-self-supervised-positive-sampling-for","title":"SSPS: Self-Supervised Positive Sampling for Robust Self-Supervised Speaker Verification","date":"2025-05-20","arxiv_id":"2505.14561","n_code_links":1,"syntology":null},{"paper":null,"title":"Representation Learning via Non-Contrastive Mutual Information","date":"2025-04-23","arxiv_id":"2504.16667","n_code_links":0,"syntology":null},{"paper":null,"title":"Impact of Language Guidance: A Reproducibility Study","date":"2025-04-10","arxiv_id":"2504.08140","n_code_links":0,"syntology":null},{"paper":null,"title":"Unsupervised Detection of Fraudulent Transactions in E-commerce Using Contrastive Learning","date":"2025-03-24","arxiv_id":"2503.18841","n_code_links":0,"syntology":null},{"paper":null,"title":"A Statistical Theory of Contrastive Learning via Approximate Sufficient Statistics","date":"2025-03-21","arxiv_id":"2503.17538","n_code_links":0,"syntology":null},{"paper":null,"title":"SinSim: Sinkhorn-Regularized SimCLR","date":"2025-02-13","arxiv_id":"2502.10478","n_code_links":0,"syntology":null},{"paper":"/paper/dataset-ownership-verification-in-contrastive","title":"Dataset Ownership Verification in Contrastive Pre-trained Models","date":"2025-02-11","arxiv_id":"2502.07276","n_code_links":1,"syntology":null},{"paper":"/paper/self-supervised-frameworks-for-speaker","title":"Self-Supervised Frameworks for Speaker Verification via Bootstrapped Positive Sampling","date":"2025-01-29","arxiv_id":"2501.17772","n_code_links":1,"syntology":null},{"paper":"/paper/enhancing-contrastive-learning-inspired-by","title":"Enhancing Contrastive Learning Inspired by the Philosophy of \"The Blind Men and the Elephant\"","date":"2024-12-21","arxiv_id":"2412.16522","n_code_links":1,"syntology":null},{"paper":null,"title":"Self-Supervised Radiograph Anatomical Region Classification -- How Clean Is Your Real-World Data?","date":"2024-12-20","arxiv_id":"2412.15967","n_code_links":0,"syntology":null},{"paper":null,"title":"Maximising Histopathology Segmentation using Minimal Labels via Self-Supervision","date":"2024-12-19","arxiv_id":"2412.15389","n_code_links":0,"syntology":null},{"paper":"/paper/amclr-unified-augmented-learning-for-cross","title":"AmCLR: Unified Augmented Learning for Cross-Modal Representations","date":"2024-12-10","arxiv_id":"2412.07979","n_code_links":1,"syntology":null},{"paper":null,"title":"Mitigating Instance-Dependent Label Noise: Integrating Self-Supervised Pretraining with Pseudo-Label Refinement","date":"2024-12-06","arxiv_id":"2412.04898","n_code_links":0,"syntology":null},{"paper":"/paper/tight-pac-bayesian-risk-certificates-for","title":"Tight PAC-Bayesian Risk Certificates for Contrastive Learning","date":"2024-12-04","arxiv_id":"2412.03486","n_code_links":1,"syntology":null},{"paper":null,"title":"Explorations in Self-Supervised Learning: Dataset Composition Testing for Object Classification","date":"2024-12-01","arxiv_id":"2412.00770","n_code_links":0,"syntology":null},{"paper":"/paper/breccia-and-basalt-classification-of-thin","title":"Breccia and basalt classification of thin sections of Apollo rocks with deep learning","date":"2024-10-28","arxiv_id":"2410.21024","n_code_links":1,"syntology":null},{"paper":null,"title":"Accelerating Augmentation Invariance Pretraining","date":"2024-10-27","arxiv_id":"2410.22364","n_code_links":0,"syntology":null},{"paper":null,"title":"Rethinking Positive Pairs in Contrastive Learning","date":"2024-10-23","arxiv_id":"2410.18200","n_code_links":0,"syntology":null},{"paper":null,"title":"SigCLR: Sigmoid Contrastive Learning of Visual Representations","date":"2024-10-22","arxiv_id":"2410.17427","n_code_links":0,"syntology":null},{"paper":null,"title":"Self-Supervised Anomaly Detection in the Wild: Favor Joint Embeddings Methods","date":"2024-10-05","arxiv_id":"2410.04289","n_code_links":0,"syntology":null},{"paper":null,"title":"BiSSL: Enhancing the Alignment Between Self-Supervised Pretraining and Downstream Fine-Tuning via Bilevel Optimization","date":"2024-10-03","arxiv_id":"2410.02387","n_code_links":0,"syntology":null},{"paper":null,"title":"Understanding the Benefits of SimCLR Pre-Training in Two-Layer Convolutional Neural Networks","date":"2024-09-27","arxiv_id":"2409.18685","n_code_links":0,"syntology":null},{"paper":"/paper/self-supervised-pretraining-for-1","title":"Self-supervised Pretraining for Cardiovascular Magnetic Resonance Cine Segmentation","date":"2024-09-26","arxiv_id":"2409.18100","n_code_links":1,"syntology":null},{"paper":"/paper/cross-model-cross-stream-learning-for-self","title":"Cross-Model Cross-Stream Learning for Self-Supervised Human Action Recognition","date":"2024-09-23","arxiv_id":null,"n_code_links":1,"syntology":null},{"paper":null,"title":"Contrastive Learning for Character Detection in Ancient Greek Papyri","date":"2024-09-16","arxiv_id":"2409.10156","n_code_links":0,"syntology":null},{"paper":null,"title":"Label-free Monitoring of Self-Supervised Learning Progress","date":"2024-09-10","arxiv_id":"2409.06612","n_code_links":0,"syntology":null},{"paper":"/paper/learning-tree-structured-composition-of-data","title":"Learning Tree-Structured Composition of Data Augmentation","date":"2024-08-26","arxiv_id":"2408.14381","n_code_links":1,"syntology":null}],"papers_shown":30,"tasks":[{"task":"/task/self-supervised-learning","name":"Self-Supervised Learning","papers":120},{"task":"/task/contrastive-learning","name":"Contrastive Learning","papers":116},{"task":"/task/representation-learning","name":"Representation Learning","papers":79},{"task":"/task/data-augmentation","name":"Data Augmentation","papers":29},{"task":"/task/image-classification","name":"Image Classification","papers":22},{"task":"/task/image-classification","name":"image-classification","papers":16},{"task":"/task/transfer-learning","name":"Transfer Learning","papers":15},{"task":"/task/linear-evaluation","name":"Linear evaluation","papers":11},{"task":"/task/object-detection","name":"Object Detection","papers":10},{"task":"/task/semantic-segmentation","name":"Semantic Segmentation","papers":10},{"task":"/task/object-detection-1","name":"object-detection","papers":10},{"task":"/task/classification-1","name":"Classification","papers":8},{"task":"/task/retrieval","name":"Retrieval","papers":8},{"task":"/task/activity-recognition","name":"Activity Recognition","papers":7},{"task":"/task/clustering","name":"Clustering","papers":7},{"task":"/task/classification","name":"General Classification","papers":7},{"task":"/task/human-activity-recognition","name":"Human Activity Recognition","papers":7},{"task":"/task/segmentation","name":"Segmentation","papers":7},{"task":"/task/benchmarking","name":"Benchmarking","papers":5},{"task":"/task/image-retrieval","name":"Image Retrieval","papers":5}],"tasks_shown":20,"n_tasks":202,"usage_by_year":[{"year":"2020","papers":25},{"year":"2021","papers":51},{"year":"2022","papers":54},{"year":"2023","papers":53},{"year":"2024","papers":46},{"year":"2025","papers":11}],"row_source":"embedded","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/simclr"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}