{"id":"arxiv:2006.09882","title":"Unsupervised Learning of Visual Features by Contrasting Cluster Assignments","date":"2020-06-17","section":"Self-supervised learning and vision","quality":0.412639,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/","source_url":"https://arxiv.org/abs/2006.09882","text_available":true,"authors":["Mathilde Caron","Ishan Misra","Julien Mairal","Priya Goyal"],"outgoing":[],"incoming":[{"source":"arxiv:2104.14294","target":"arxiv:2006.09882","evidence":"title","id":"arxiv:2104.14294","title":"Emerging Properties in Self-Supervised Vision Transformers","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/4372186d141ca514b08e93bb3b1e26fdbcba9ce62187fb772c14bf223160c06d/"},{"source":"arxiv:2105.04906","target":"arxiv:2006.09882","evidence":"title","id":"arxiv:2105.04906","title":"VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/822b6414fb02bd6c746e0d559fd3dcc60766476142ebae7795b02bc6c45223ce/"},{"source":"arxiv:2111.07832","target":"arxiv:2006.09882","evidence":"title","id":"arxiv:2111.07832","title":"iBOT: Image BERT Pre-Training with Online Tokenizer","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/2bd0269c0872fc518db65c2ca98dd30b5d38701f79b0a72fec2ed49648ba589f/"},{"source":"arxiv:2301.08243","target":"arxiv:2006.09882","evidence":"id","id":"arxiv:2301.08243","title":"Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/475bac0a3b29edd42642bbe10d1a41bfb18a3022ac95b9b1a3db9e5d5d0b95fe/"},{"source":"arxiv:2304.07193","target":"arxiv:2006.09882","evidence":"title","id":"arxiv:2304.07193","title":"DINOv2: Learning Robust Visual Features without Supervision","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/b15933596df35497024556ae88bfa696f3b0e8372b678194f04f1bfb9f3be553/"},{"source":"arxiv:2304.09355","target":"arxiv:2006.09882","evidence":"title","id":"arxiv:2304.09355","title":"To Compress or Not to Compress- Self-Supervised Learning and Information Theory: A Review","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6de01284dffdeadbd92020fc614ff702f27bc9c49215a1658ea3ca8a4a0c36ce/"},{"source":"arxiv:2304.12210","target":"arxiv:2006.09882","evidence":"title","id":"arxiv:2304.12210","title":"A Cookbook of Self-Supervised Learning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/518a026819227af3e1838911bfadf26c1515786a3af305de5c2e77efb052ecfd/"},{"source":"arxiv:2311.06242","target":"arxiv:2006.09882","evidence":"title","id":"arxiv:2311.06242","title":"Florence-2: Advancing a Unified Representation for a Variety of Vision Tasks","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/edc2679fff98a10c70395e79c58b56eaf9ec0001691dbca2b33ba94a401a8398/"},{"source":"arxiv:2312.17742","target":"arxiv:2006.09882","evidence":"title","id":"arxiv:2312.17742","title":"Learning Vision from Models Rivals Learning Vision from Data","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/71dc1683019904602d643d16348a76b98c556b924ca964e3949d8d6420dc4402/"},{"source":"arxiv:2606.02572","target":"arxiv:2006.09882","evidence":"title","id":"arxiv:2606.02572","title":"VISReg: Variance-Invariance-Sketching Regularization for JEPA training","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/262c1938984761f99c55d9ae1237e686b2d6ab94b69d7181d11dceab2a7210fb/"}],"text_parts":[{"start":0,"end":14620,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html"},{"start":14620,"end":29877,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html"},{"start":29877,"end":32783,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-3.html"}],"contents":[{"title":"Unsupervised Learning of Visual Features by Contrasting Cluster Assignments","offset":0,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"Abstract","offset":79,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"1 Introduction","offset":1602,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"2 Related Work","offset":6630,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"Instance and contrastive learning.","offset":6649,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"Clustering for deep representation learning.","offset":7693,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"Handcrafted pretext tasks.","offset":8639,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"3 Method","offset":9378,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"3.1 Online clustering","offset":11828,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"Swapped prediction problem.","offset":12762,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"Computing codes online.","offset":14589,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-1.html#text"},{"title":"Working with small batches.","offset":18551,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"3.2 Multi-crop: Augmenting views with smaller images","offset":19287,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"4 Main Results","offset":20713,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"4.1 Evaluating the unsupervised features on ImageNet","offset":21064,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"4.2 Transferring unsupervised features to downstream tasks","offset":22434,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"4.3 Training with small batches","offset":24223,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"5 Ablation Study","offset":25644,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"5.1 Clustering-based self-supervised learning","offset":25665,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"Improving prior clustering-based approaches.","offset":25716,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"Comparing clustering with contrastive instance learning.","offset":27103,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"Advantage of SwAV compared to DeepCluster-v2.","offset":27629,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"5.2 Applying the multi-crop strategy to different methods","offset":28693,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"5.3 Impact of longer training","offset":29290,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"5.4 Unsupervised pretraining on a large uncurated dataset","offset":29814,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-2.html#text"},{"title":"6 Discussion","offset":31827,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-3.html#text"},{"title":"Acknowledgement.","offset":32432,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/part-3.html#text"}],"text_code_points":32783,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6601018611a96dff6f5324f2e4f39f62484917cf4d8f8a8d5baa7f4be1deb433/source.md"}