{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/data-feedback-loops-model-driven","title":"Data Feedback Loops: Model-driven Amplification of Dataset Biases","arxiv_id":"2209.03942","date":"2022-09-08","proceeding":null,"authors":["Rohan Taori","Tatsunori B. Hashimoto"],"abstract":"Datasets scraped from the internet have been critical to the successes of large-scale machine learning. Yet, this very success puts the utility of future internet-derived datasets at potential risk, as model outputs begin to replace human annotations as a source of supervision. In this work, we first formalize a system where interactions with one model are recorded as history and scraped as training data in the future. We then analyze its stability over time by tracking changes to a test-time bias statistic (e.g. gender bias of model predictions). We find that the degree of bias amplification is closely linked to whether the model's outputs behave like samples from the training distribution, a behavior which we characterize and define as consistent calibration. Experiments in three conditional prediction scenarios - image classification, visual role-labeling, and language generation - demonstrate that models that exhibit a sampling-like behavior are more calibrated and thus more stable. Based on this insight, we propose an intervention to help calibrate and stabilize unstable feedback systems. Code is available at https://github.com/rtaori/data_feedback.","url_abs":"https://arxiv.org/abs/2209.03942v1","url_pdf":"https://arxiv.org/pdf/2209.03942v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"data-feedback-loops-model-driven","repo_url":"https://github.com/rtaori/data_feedback","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"text-generation","task_name":"Text Generation"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2209.03942","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2209.03942"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/rtaori/data_feedback","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"unverified":17},"by_repo_kind":{"official":{"samples":17,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"98c5a9335415160f","entry":"batch_norm","repo":"rtaori/data_feedback","repo_kind":"official","path":"image_classification/baidunet.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/image_classification/baidunet.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"98c5a9335415160f"}},{"code_sha256_prefix":"faa2eeb8672f4711","entry":"collapse_annotations","repo":"rtaori/data_feedback","repo_kind":"official","path":"visual_role_labeling/data.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/visual_role_labeling/data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"faa2eeb8672f4711"}},{"code_sha256_prefix":"ac25750813c3e6f9","entry":"conv_bn","repo":"rtaori/data_feedback","repo_kind":"official","path":"image_classification/baidunet.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/image_classification/baidunet.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ac25750813c3e6f9"}},{"code_sha256_prefix":"81d473f07388d56a","entry":"format_dict","repo":"rtaori/data_feedback","repo_kind":"official","path":"visual_role_labeling/baseline_crf.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/visual_role_labeling/baseline_crf.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"81d473f07388d56a"}},{"code_sha256_prefix":"6d38eeb275e78417","entry":"get_bias_std_lines","repo":"rtaori/data_feedback","repo_kind":"official","path":"plotting/plotting_utils.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/plotting/plotting_utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"6d38eeb275e78417"}},{"code_sha256_prefix":"819ba6bdeceaaf30","entry":"load_cifar5m","repo":"rtaori/data_feedback","repo_kind":"official","path":"image_classification/data.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/image_classification/data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"819ba6bdeceaaf30"}},{"code_sha256_prefix":"804d4cfbb5a899a0","entry":"load_cinic10","repo":"rtaori/data_feedback","repo_kind":"official","path":"image_classification/data.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/image_classification/data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"804d4cfbb5a899a0"}},{"code_sha256_prefix":"dfae76f2b889026e","entry":"load_data","repo":"rtaori/data_feedback","repo_kind":"official","path":"image_classification/data.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/image_classification/data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"dfae76f2b889026e"}},{"code_sha256_prefix":"935670c8c277b699","entry":"load_rtp_dataset","repo":"rtaori/data_feedback","repo_kind":"official","path":"language_generation/data.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/language_generation/data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"935670c8c277b699"}},{"code_sha256_prefix":"71cfaeb6fde82196","entry":"rand_split_dataset","repo":"rtaori/data_feedback","repo_kind":"official","path":"visual_role_labeling/data.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/visual_role_labeling/data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"71cfaeb6fde82196"}},{"code_sha256_prefix":"cc062ce9df817b3a","entry":"rand_split_test_set","repo":"rtaori/data_feedback","repo_kind":"official","path":"visual_role_labeling/data.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/visual_role_labeling/data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"cc062ce9df817b3a"}},{"code_sha256_prefix":"dc51887b30539963","entry":"residual","repo":"rtaori/data_feedback","repo_kind":"official","path":"image_classification/baidunet.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/image_classification/baidunet.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"dc51887b30539963"}},{"code_sha256_prefix":"9a0a2bca8191c1b6","entry":"shuffle_tensor","repo":"rtaori/data_feedback","repo_kind":"official","path":"image_classification/utils.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/image_classification/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"9a0a2bca8191c1b6"}},{"code_sha256_prefix":"116d1355faa1c187","entry":"theorem1_bound","repo":"rtaori/data_feedback","repo_kind":"official","path":"plotting/feedback_bound.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/plotting/feedback_bound.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"116d1355faa1c187"}},{"code_sha256_prefix":"7e564a313204ad07","entry":"theorem1_continual_bound","repo":"rtaori/data_feedback","repo_kind":"official","path":"plotting/feedback_bound.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/plotting/feedback_bound.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"7e564a313204ad07"}},{"code_sha256_prefix":"997469cb5a9c5f62","entry":"tokenize_generation","repo":"rtaori/data_feedback","repo_kind":"official","path":"language_generation/data.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/language_generation/data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"997469cb5a9c5f62"}},{"code_sha256_prefix":"a5dc158924c783cc","entry":"tokenize_training","repo":"rtaori/data_feedback","repo_kind":"official","path":"language_generation/data.py","file_url":"https://github.com/rtaori/data_feedback/blob/HEAD/language_generation/data.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"a5dc158924c783cc"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}