{"url":"/sota/continual-learning-on-visual-domain-decathlon","task":{"name":"Continual Learning","url":"/task/continual-learning","note":null},"dataset":{"name":"visual domain decathlon (10 tasks)","url":"/dataset/visual-domain-decathlon"},"category":"Computer Vision","categories":["Computer Vision","Methodology","Natural Language Processing"],"category_note":null,"description":"**Continual Learning** (also known as **Incremental Learning**, **Life-long Learning**) is a concept to learn a model for a large number of tasks sequentially without forgetting knowledge obtained from the preceding tasks, where the data in the old tasks are not available anymore during training new ones.  \r\nIf not mentioned, the benchmarks here are **Task-CL**, where task-id is provided on validation.\r\n\r\nSource:  \r\n[Continual Learning by Asymmetric Loss Approximation with Single-Side Overestimation](https://arxiv.org/abs/1908.02984)  \r\n[Three scenarios for continual learning](https://arxiv.org/abs/1904.07734)  \r\n[Lifelong Machine Learning](https://books.google.ca/books/about/Lifelong_Machine_Learning.html?id=JQ5pDwAAQBAJ&redir_esc=y)  \r\n[Continual lifelong learning with neural networks: A review](https://www.sciencedirect.com/science/article/pii/S0893608019300231)","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["decathlon discipline (Score)","Avg. Accuracy"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"decathlon discipline (Score)":"higher","Avg. Accuracy":"higher"}},"counts":{"rows":14,"rows_with_code":12,"rows_with_paper_page":14,"rows_dated":14,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"NetTailor","metrics":{"Avg. Accuracy":"79.64","decathlon discipline (Score)":"3744"},"uses_additional_data":false,"paper_date":"2019-06-29","paper":"/paper/nettailor-tuning-the-architecture-not-just-1","paper_url":"https://arxiv.org/abs/1907.00274v1","paper_title":"NetTailor: Tuning the Architecture, Not Just the Weights","code":"https://github.com/pedro-morgado/nettailor","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"Depthwise Soft Sharing","metrics":{"decathlon discipline (Score)":"3507"},"uses_additional_data":false,"paper_date":"2019-02-03","paper":"/paper/depthwise-convolution-is-all-you-need-for","paper_url":"http://arxiv.org/abs/1902.00927v2","paper_title":"Depthwise Convolution is All You Need for Learning Multiple Visual Domains","code":"https://github.com/yunhuiguo/Depthwise_Convolution_for_Multiple_Domain_Learning","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"Parallel Res. adapt.","metrics":{"Avg. Accuracy":"78.07","decathlon discipline (Score)":"3412"},"uses_additional_data":false,"paper_date":"2018-03-27","paper":"/paper/efficient-parametrization-of-multi-domain","paper_url":"http://arxiv.org/abs/1803.10082v1","paper_title":"Efficient parametrization of multi-domain deep neural networks","code":"https://github.com/srebuffi/residual_adapters","n_code_links":3,"syntology":{"n_ran":2,"n_unverified":1,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":4,"model":"Depthwise Sharing","metrics":{"decathlon discipline (Score)":"3234"},"uses_additional_data":false,"paper_date":"2019-02-03","paper":"/paper/depthwise-convolution-is-all-you-need-for","paper_url":"http://arxiv.org/abs/1902.00927v2","paper_title":"Depthwise Convolution is All You Need for Learning Multiple Visual Domains","code":"https://github.com/yunhuiguo/Depthwise_Convolution_for_Multiple_Domain_Learning","n_code_links":1,"syntology":null},{"rank_in_archive_order":5,"model":"Series Res. adapt.","metrics":{"decathlon discipline (Score)":"3159"},"uses_additional_data":false,"paper_date":"2018-03-27","paper":"/paper/efficient-parametrization-of-multi-domain","paper_url":"http://arxiv.org/abs/1803.10082v1","paper_title":"Efficient parametrization of multi-domain deep neural networks","code":"https://github.com/srebuffi/residual_adapters","n_code_links":3,"syntology":{"n_ran":2,"n_unverified":1,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":6,"model":"Res. adapt. (large)","metrics":{"decathlon discipline (Score)":"3131"},"uses_additional_data":false,"paper_date":"2017-05-22","paper":"/paper/learning-multiple-visual-domains-with","paper_url":"http://arxiv.org/abs/1705.08045v5","paper_title":"Learning multiple visual domains with residual adapters","code":"https://github.com/srebuffi/residual_adapters","n_code_links":2,"syntology":null},{"rank_in_archive_order":7,"model":"DAN","metrics":{"Avg. Accuracy":"77.01","decathlon discipline (Score)":"2851"},"uses_additional_data":false,"paper_date":"2017-05-11","paper":"/paper/incremental-learning-through-deep-adaptation","paper_url":"http://arxiv.org/abs/1705.04228v2","paper_title":"Incremental Learning Through Deep Adaptation","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":8,"model":"Piggyback","metrics":{"Avg. Accuracy":"76.60","decathlon discipline (Score)":"2838"},"uses_additional_data":false,"paper_date":"2018-01-19","paper":"/paper/piggyback-adapting-a-single-network-to","paper_url":"http://arxiv.org/abs/1801.06519v2","paper_title":"Piggyback: Adapting a Single Network to Multiple Tasks by Learning to Mask Weights","code":"https://github.com/arunmallya/piggyback","n_code_links":1,"syntology":null},{"rank_in_archive_order":9,"model":"Res. adapt. finetune all","metrics":{"decathlon discipline (Score)":"2643"},"uses_additional_data":false,"paper_date":"2017-05-22","paper":"/paper/learning-multiple-visual-domains-with","paper_url":"http://arxiv.org/abs/1705.08045v5","paper_title":"Learning multiple visual domains with residual adapters","code":"https://github.com/srebuffi/residual_adapters","n_code_links":2,"syntology":null},{"rank_in_archive_order":10,"model":"Res. adapt. decay","metrics":{"decathlon discipline (Score)":"2621"},"uses_additional_data":false,"paper_date":"2017-05-22","paper":"/paper/learning-multiple-visual-domains-with","paper_url":"http://arxiv.org/abs/1705.08045v5","paper_title":"Learning multiple visual domains with residual adapters","code":"https://github.com/srebuffi/residual_adapters","n_code_links":2,"syntology":null},{"rank_in_archive_order":11,"model":"LwF","metrics":{"Avg. Accuracy":"76.93","decathlon discipline (Score)":"2515"},"uses_additional_data":false,"paper_date":"2016-06-29","paper":"/paper/learning-without-forgetting","paper_url":"http://arxiv.org/abs/1606.09282v3","paper_title":"Learning without Forgetting","code":"https://github.com/ContinualAI/avalanche","n_code_links":12,"syntology":{"n_ran":9,"n_unverified":6,"n_samples":15,"n_pointer_only_licence":1}},{"rank_in_archive_order":12,"model":"Res. adapt. dom-pred","metrics":{"decathlon discipline (Score)":"2503"},"uses_additional_data":false,"paper_date":"2017-05-22","paper":"/paper/learning-multiple-visual-domains-with","paper_url":"http://arxiv.org/abs/1705.08045v5","paper_title":"Learning multiple visual domains with residual adapters","code":"https://github.com/srebuffi/residual_adapters","n_code_links":2,"syntology":null},{"rank_in_archive_order":13,"model":"Res. adapt.","metrics":{"decathlon discipline (Score)":"2118"},"uses_additional_data":false,"paper_date":"2017-05-22","paper":"/paper/learning-multiple-visual-domains-with","paper_url":"http://arxiv.org/abs/1705.08045v5","paper_title":"Learning multiple visual domains with residual adapters","code":"https://github.com/srebuffi/residual_adapters","n_code_links":2,"syntology":null},{"rank_in_archive_order":14,"model":"BN adapt.","metrics":{"decathlon discipline (Score)":"1363"},"uses_additional_data":false,"paper_date":"2017-01-25","paper":"/paper/universal-representationsthe-missing-link","paper_url":"http://arxiv.org/abs/1701.07275v1","paper_title":"Universal representations:The missing link between faces, text, planktons, and cat breeds","code":null,"n_code_links":0,"syntology":null}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,885 of the 9,623 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9623,"papers_checked":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2737},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":3,"rows_with_any_sample_ran":3,"distinct_papers_with_graph_line":2,"distinct_papers_with_any_sample_ran":2,"samples_over_distinct_papers":{"n_ran":11,"n_unverified":7,"n_samples":18,"n_pointer_only_licence":1,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":13,"n_unverified":8,"n_samples":21,"n_pointer_only_licence":1,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}