{"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/effort-efficient-orthogonal-modeling-for","title":"Orthogonal Subspace Decomposition for Generalizable AI-Generated Image Detection","arxiv_id":"2411.15633","date":"2024-11-23","proceeding":null,"authors":["Zhiyuan Yan","Jiangming Wang","Peng Jin","Ke-Yue Zhang","Chengchun Liu","Shen Chen","Taiping Yao","Shouhong Ding","Baoyuan Wu","Li Yuan"],"abstract":"AI-generated images (AIGIs), such as natural or face images, have become increasingly important yet challenging. In this paper, we start from a new perspective to excavate the reason behind the failure generalization in AIGI detection, named the \\textit{asymmetry phenomenon}, where a naively trained detector tends to favor overfitting to the limited and monotonous fake patterns, causing the feature space to become highly constrained and low-ranked, which is proved seriously limiting the expressivity and generalization. One potential remedy is incorporating the pre-trained knowledge within the vision foundation models (higher-ranked) to expand the feature space, alleviating the model's overfitting to fake. To this end, we employ Singular Value Decomposition (SVD) to decompose the original feature space into \\textit{two orthogonal subspaces}. By freezing the principal components and adapting only the remained components, we preserve the pre-trained knowledge while learning fake patterns. Compared to existing full-parameters and LoRA-based tuning methods, we explicitly ensure orthogonality, enabling the higher rank of the whole feature space, effectively minimizing overfitting and enhancing generalization. We finally identify a crucial insight: our method implicitly learns \\textit{a vital prior that fakes are actually derived from the real}, indicating a hierarchical relationship rather than independence. Modeling this prior, we believe, is essential for achieving superior generalization. Our codes are publicly available at \\href{https://github.com/YZY-stack/Effort-AIGI-Detection}{GitHub}.","url_abs":"https://arxiv.org/abs/2411.15633v4","url_pdf":"https://arxiv.org/pdf/2411.15633v4.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":"effort-efficient-orthogonal-modeling-for","repo_url":"https://github.com/yzy-stack/effort-aigi-detection","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"effort-efficient-orthogonal-modeling-for","repo_url":"https://github.com/sclbd/deepfakebench","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"face-swapping","task_name":"Face Swapping"},{"task_slug":"synthetic-image-detection","task_name":"Synthetic Image Detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2411.15633","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2411.15633"}},"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/yzy-stack/effort-aigi-detection","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/sclbd/deepfakebench","reach":{"status":"ok","spdx":"NOASSERTION"}}],"summary":{"ran_draft_wrong":2,"ran_violates":1,"unverified":17},"by_repo_kind":{"official":{"samples":11,"ran":3,"repositories":1},"listed":{"samples":9,"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":20,"samples":[{"code_sha256_prefix":"6e51f7833599876f","entry":"choose_metric","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"DeepfakeBench/analysis/logits_decision_boundary.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/DeepfakeBench/analysis/logits_decision_boundary.py","link_basis":"plan_row","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"6e51f7833599876f"}},{"code_sha256_prefix":"4139aa85d28b1ed5","entry":"find_best_threshold","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"UniversalFakeDetect_Benchmark/validate.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/UniversalFakeDetect_Benchmark/validate.py","link_basis":"plan_row","language":"python","status":"ran_violates","verification_level":1,"contract_check":"VIOLATES","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"4139aa85d28b1ed5"}},{"code_sha256_prefix":"8501ef758e4a9824","entry":"png2jpg","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"UniversalFakeDetect_Benchmark/validate.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/UniversalFakeDetect_Benchmark/validate.py","link_basis":"plan_row","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"8501ef758e4a9824"}},{"code_sha256_prefix":"9132d6fa35e95be4","entry":"apply_svd_residual_to_self_attn","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"DeepfakeBench/training/detectors/effort_detector.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/DeepfakeBench/training/detectors/effort_detector.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"9132d6fa35e95be4"}},{"code_sha256_prefix":"f3bfc8d27674781c","entry":"apply_svd_residual_to_self_attn","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"UniversalFakeDetect_Benchmark/models/clip_models.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/UniversalFakeDetect_Benchmark/models/clip_models.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"f3bfc8d27674781c"}},{"code_sha256_prefix":"c492630b402afb1e","entry":"calculate_average_frequency_spectra","repo":"sclbd/deepfakebench","repo_kind":"listed","path":"analysis/frequency.py","file_url":"https://github.com/sclbd/deepfakebench/blob/HEAD/analysis/frequency.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"c492630b402afb1e"}},{"code_sha256_prefix":"db6cc0c7755a186f","entry":"calculate_logits_and_confidence","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"DeepfakeBench/analysis/logits_decision_boundary.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/DeepfakeBench/analysis/logits_decision_boundary.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"db6cc0c7755a186f"}},{"code_sha256_prefix":"3b8c0e8be57b8ce7","entry":"create_logger","repo":"sclbd/deepfakebench","repo_kind":"listed","path":"analysis/frequency.py","file_url":"https://github.com/sclbd/deepfakebench/blob/HEAD/analysis/frequency.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"3b8c0e8be57b8ce7"}},{"code_sha256_prefix":"5b93d8d118f8fe72","entry":"create_logger","repo":"sclbd/deepfakebench","repo_kind":"listed","path":"preprocessing/preprocess.py","file_url":"https://github.com/sclbd/deepfakebench/blob/HEAD/preprocessing/preprocess.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"5b93d8d118f8fe72"}},{"code_sha256_prefix":"a06ab489f0d55d2c","entry":"create_logger","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"DeepfakeBench/training/logger.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/DeepfakeBench/training/logger.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"a06ab489f0d55d2c"}},{"code_sha256_prefix":"e8437cbb382c1c2d","entry":"get_keypts","repo":"sclbd/deepfakebench","repo_kind":"listed","path":"preprocessing/preprocess.py","file_url":"https://github.com/sclbd/deepfakebench/blob/HEAD/preprocessing/preprocess.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"e8437cbb382c1c2d"}},{"code_sha256_prefix":"e839d6953f748e55","entry":"get_top3_avg_auc","repo":"sclbd/deepfakebench","repo_kind":"listed","path":"analysis/model_archi.py","file_url":"https://github.com/sclbd/deepfakebench/blob/HEAD/analysis/model_archi.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"e839d6953f748e55"}},{"code_sha256_prefix":"23ec693b6f2f66bd","entry":"load_images_from_paths","repo":"sclbd/deepfakebench","repo_kind":"listed","path":"analysis/frequency.py","file_url":"https://github.com/sclbd/deepfakebench/blob/HEAD/analysis/frequency.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"23ec693b6f2f66bd"}},{"code_sha256_prefix":"3f6fd132ed3115aa","entry":"replace_name","repo":"sclbd/deepfakebench","repo_kind":"listed","path":"analysis/auc_fromaug.py","file_url":"https://github.com/sclbd/deepfakebench/blob/HEAD/analysis/auc_fromaug.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"3f6fd132ed3115aa"}},{"code_sha256_prefix":"6344adb109654c9d","entry":"replace_name","repo":"sclbd/deepfakebench","repo_kind":"listed","path":"analysis/auc_table1_fromrecord.py","file_url":"https://github.com/sclbd/deepfakebench/blob/HEAD/analysis/auc_table1_fromrecord.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"6344adb109654c9d"}},{"code_sha256_prefix":"33f6e13013ae471c","entry":"replace_name","repo":"sclbd/deepfakebench","repo_kind":"listed","path":"analysis/auc_table2_fromrecord.py","file_url":"https://github.com/sclbd/deepfakebench/blob/HEAD/analysis/auc_table2_fromrecord.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"33f6e13013ae471c"}},{"code_sha256_prefix":"034042e10ded5933","entry":"replace_with_svd_residual","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"DeepfakeBench/training/detectors/effort_detector.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/DeepfakeBench/training/detectors/effort_detector.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"034042e10ded5933"}},{"code_sha256_prefix":"4ffd15e35234031a","entry":"replace_with_svd_residual","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"UniversalFakeDetect_Benchmark/models/clip_models.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/UniversalFakeDetect_Benchmark/models/clip_models.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"4ffd15e35234031a"}},{"code_sha256_prefix":"751cd91f7274f805","entry":"translate_duplicate","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"UniversalFakeDetect_Benchmark/validate.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/UniversalFakeDetect_Benchmark/validate.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"751cd91f7274f805"}},{"code_sha256_prefix":"7742460f647f9ca2","entry":"unnormalize","repo":"yzy-stack/effort-aigi-detection","repo_kind":"official","path":"UniversalFakeDetect_Benchmark/util.py","file_url":"https://github.com/yzy-stack/effort-aigi-detection/blob/HEAD/UniversalFakeDetect_Benchmark/util.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"7742460f647f9ca2"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}