{"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/decorrelated-adversarial-learning-for-age","title":"Decorrelated Adversarial Learning for Age-Invariant Face Recognition","arxiv_id":"1904.04972","date":"2019-04-10","proceeding":"CVPR 2019 6","authors":["Hao Wang","Dihong Gong","Zhifeng Li","Wei Liu"],"abstract":"There has been an increasing research interest in age-invariant face\nrecognition. However, matching faces with big age gaps remains a challenging\nproblem, primarily due to the significant discrepancy of face appearances\ncaused by aging. To reduce such a discrepancy, in this paper we propose a novel\nalgorithm to remove age-related components from features mixed with both\nidentity and age information. Specifically, we factorize a mixed face feature\ninto two uncorrelated components: identity-dependent component and\nage-dependent component, where the identity-dependent component includes\ninformation that is useful for face recognition. To implement this idea, we\npropose the Decorrelated Adversarial Learning (DAL) algorithm, where a\nCanonical Mapping Module (CMM) is introduced to find the maximum correlation\nbetween the paired features generated by a backbone network, while the backbone\nnetwork and the factorization module are trained to generate features reducing\nthe correlation. Thus, the proposed model learns the decomposed features of age\nand identity whose correlation is significantly reduced. Simultaneously, the\nidentity-dependent feature and the age-dependent feature are respectively\nsupervised by ID and age preserving signals to ensure that they both contain\nthe correct information. Extensive experiments are conducted on popular\npublic-domain face aging datasets (FG-NET, MORPH Album 2, and CACD-VS) to\ndemonstrate the effectiveness of the proposed approach.","url_abs":"http://arxiv.org/abs/1904.04972v1","url_pdf":"http://arxiv.org/pdf/1904.04972v1.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":"decorrelated-adversarial-learning-for-age","repo_url":"https://github.com/neverUseThisName/Decorrelated-Adversarial-Learning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"age-invariant-face-recognition","task_name":"Age-Invariant Face Recognition"},{"task_slug":"face-recognition","task_name":"Face Recognition"},{"task_slug":"morph","task_name":"MORPH"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/age-invariant-face-recognition-on-cacdvs","task":"Age-Invariant Face Recognition","dataset":"CACDVS","model":"DAL","rank_in_archive_order":3,"of":9,"metrics":{"Accuracy":"99.4%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.04972","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}