{"id":26055278,"url":"https://github.com/takashiishida/comp","last_synced_at":"2025-10-14T04:27:10.670Z","repository":{"id":71031441,"uuid":"185098926","full_name":"takashiishida/comp","owner":"takashiishida","description":"[NeurIPS 2017] [ICML 2019] Code for complementary-label learning","archived":false,"fork":false,"pushed_at":"2024-01-21T09:51:32.000Z","size":3370,"stargazers_count":48,"open_issues_count":0,"forks_count":17,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-04-11T02:56:52.638Z","etag":null,"topics":["deep-learning","machine-learning"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/takashiishida.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2019-05-06T00:40:59.000Z","updated_at":"2025-04-05T01:54:17.000Z","dependencies_parsed_at":"2025-03-08T10:11:10.875Z","dependency_job_id":null,"html_url":"https://github.com/takashiishida/comp","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/takashiishida/comp","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/takashiishida%2Fcomp","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/takashiishida%2Fcomp/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/takashiishida%2Fcomp/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/takashiishida%2Fcomp/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/takashiishida","download_url":"https://codeload.github.com/takashiishida/comp/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/takashiishida%2Fcomp/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":279017947,"owners_count":26086213,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-10-14T02:00:06.444Z","response_time":60,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-learning","machine-learning"],"created_at":"2025-03-08T10:00:57.259Z","updated_at":"2025-10-14T04:27:10.654Z","avatar_url":"https://github.com/takashiishida.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Complementary-Label Learning\n\nThis repository gives the implementation for *complementary-label learning* from the ICML 2019 paper [1], the ECCV 2018 paper [2], and the NeurIPS 2017 paper [3].\n\n## Requirements\n- Python 3.6\n- numpy 1.14\n- PyTorch 1.1\n- torchvision 0.2\n\n## Demo\nThe following demo will show the results with the MNIST dataset.  After running the code, you should see a text file with the results saved in the same directory.  The results will have three columns: epoch number, training accuracy, and test accuracy.\n\n```bash\npython demo.py -h\n```\n\n#### Methods and models\nIn `demo.py`, specify the `method` argument to choose one of the 5 methods available:\n\n- `ga`: Gradient ascent version (Algorithm 1) in [1].\n- `nn`: Non-negative risk estimator with the max operator in [1].\n- `free`: Assumption-free risk estimator based on Theorem 1 in [1].\n- `forward`: Forward correction method in [2].\n- `pc`: Pairwise comparison with sigmoid loss in [3].\n\nSpecify the `model` argument:\n\n- `linear`: Linear model\n- `mlp`: Multi-layer perceptron with one hidden layer (500 units)\n\n## Reference\n1. T. Ishida, G. Niu, A. K. Menon, and M. Sugiyama.\u003cbr\u003e**Complementary-label learning for arbitrary losses and models**.\u003cbr\u003eIn *ICML 2019*.\u003cbr\u003e[[paper]](https://arxiv.org/abs/1810.04327)\n2. Yu, X., Liu, T., Gong, M., and Tao, D.\u003cbr\u003e**Learning with biased complementary labels**.\u003cbr\u003eIn *ECCV 2018*.\u003cbr\u003e[[paper]](https://arxiv.org/abs/1711.09535)\n3. T. Ishida, G. Niu, W. Hu, and M. Sugiyama.\u003cbr\u003e**Learning from complementary labels**.\u003cbr\u003eIn *NeurIPS 2017*.\u003cbr\u003e[[paper]](https://arxiv.org/abs/1705.07541)\n\nIf you have any further questions, please feel free to send an e-mail to: ishida at ms.k.u-tokyo.ac.jp.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ftakashiishida%2Fcomp","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Ftakashiishida%2Fcomp","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ftakashiishida%2Fcomp/lists"}