{"id":13576220,"url":"https://github.com/julilien/MitigatingLabelNoiseDataAmbiguation","last_synced_at":"2025-04-05T05:31:09.826Z","repository":{"id":168910989,"uuid":"606024209","full_name":"julilien/MitigatingLabelNoiseDataAmbiguation","owner":"julilien","description":"Supplementary material and code for \"Mitigating Label Noise through Data Ambiguation\" as published at AAAI 2024.","archived":false,"fork":false,"pushed_at":"2024-04-27T14:25:43.000Z","size":903,"stargazers_count":4,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2024-11-05T12:33:11.816Z","etag":null,"topics":["aaai","aaai2024","data-ambiguation","deep-learning","label-noise","label-relaxation","machine-learning","robustness","weakly-supervised-learning"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/julilien.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":"2023-02-24T12:31:41.000Z","updated_at":"2024-08-09T19:57:37.000Z","dependencies_parsed_at":"2024-01-16T20:04:07.933Z","dependency_job_id":"afd5bb1a-517d-4df2-a34b-01a777202152","html_url":"https://github.com/julilien/MitigatingLabelNoiseDataAmbiguation","commit_stats":null,"previous_names":["julilien/mitigatinglabelnoisedataambiguation"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/julilien%2FMitigatingLabelNoiseDataAmbiguation","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/julilien%2FMitigatingLabelNoiseDataAmbiguation/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/julilien%2FMitigatingLabelNoiseDataAmbiguation/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/julilien%2FMitigatingLabelNoiseDataAmbiguation/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/julilien","download_url":"https://codeload.github.com/julilien/MitigatingLabelNoiseDataAmbiguation/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":247294190,"owners_count":20915332,"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","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":["aaai","aaai2024","data-ambiguation","deep-learning","label-noise","label-relaxation","machine-learning","robustness","weakly-supervised-learning"],"created_at":"2024-08-01T15:01:08.103Z","updated_at":"2025-04-05T05:31:04.815Z","avatar_url":"https://github.com/julilien.png","language":"Python","funding_links":[],"categories":["Python"],"sub_categories":[],"readme":"# Mitigating Label Noise through Data Ambiguation\n\nThis repository contains an implementation of [Mitigating Label Noise through Data Ambiguation](http://arxiv.org/abs/2305.13764\n) to be presented at AAAI-24. Please cite it as follows:\n\n```\n@misc{lienen2023mitigating,\n      title={Mitigating Label Noise through Data Ambiguation}, \n      author={Julian Lienen and Eyke Hüllermeier},\n      year={2023},\n      eprint={2305.13764},\n      archivePrefix={arXiv},\n      primaryClass={cs.LG}\n}\n```\n\n## Requirements\n\nTo install all required packages, you need to run\n```\npip install -r requirements.txt\n```\n\nThe code has been tested using Python 3.9 on Ubuntu 2*.* systems. We trained our models on machines with Nvidia GPUs (we tested CUDA 10.1, 11.1 and 11.6). Furthermore, we recommend to use [Python virtual environments](https://docs.python.org/3/tutorial/venv.html) to get a clean Python environment for the execution without any dependency problems.\n\nAs a required prerequisite, the `config.ini` needs to be populated with parameters to set the output directory (`BASE_PATH`), a directory for temporary artifacts (`TMP_PATH`) and an output directory for plots (`PLOT_DIR`). \n\n## Datasets\n\nAll datasets except for CIFAR-10(0)N, WebVision and Clothing1M are downloaded automatically. Webvision is available [here](https://data.vision.ee.ethz.ch/cvl/webvision/dataset2017.html), whereas access to Clothing1M has to be [explicitly granted](https://github.com/Cysu/noisy_label) by the owner. CIFAR-10(0)is available [here](https://github.com/UCSC-REAL/cifar-10-100n). All data needs to be stored in the specified `--data_dir` given as parameter to the training script (see next section).\n\n## Training and Evaluation\n\nFor the training and evaluation, you have to call the following function (e.g., for CIFAR-10 with 25 % symmetric synthetic noise for our loss):\n\n```\nCUDA_VISIBLE_DEVICES=\u003cthe numeric ID(s) of your CUDA device(s)\u003e python train.py --dataset=cifar10  --model resnet34 --seed 0 --loss RDA --adaptive_lrvar2 --adaptive_lrvar2_start_beta 0.75 --lrvar2_beta 0.6 --adaptive_lrvar2_type cosine --lr 0.02 --decay_type cosine --label_noise 0.25\n```\n\n`--help` allows for printing out all parameter options. All results presented in the paper were computed based on the training scripts `train.py`.\n\n## License\n\nOur code uses the Apache 2.0 License, which we attached as `LICENSE` file in this repository. \n\nFeel free to re-use our code. We would be happy to see our ideas put into practice.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjulilien%2FMitigatingLabelNoiseDataAmbiguation","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fjulilien%2FMitigatingLabelNoiseDataAmbiguation","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjulilien%2FMitigatingLabelNoiseDataAmbiguation/lists"}