{"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/feedback-prop-convolutional-neural-network","title":"Feedback-prop: Convolutional Neural Network Inference under Partial Evidence","arxiv_id":"1710.08049","date":"2017-10-23","proceeding":"CVPR 2018 6","authors":["Tianlu Wang","Kota Yamaguchi","Vicente Ordonez"],"abstract":"We propose an inference procedure for deep convolutional neural networks\n(CNNs) when partial evidence is available. Our method consists of a general\nfeedback-based propagation approach (feedback-prop) that boosts the prediction\naccuracy for an arbitrary set of unknown target labels when the values for a\nnon-overlapping arbitrary set of target labels are known. We show that existing\nmodels trained in a multi-label or multi-task setting can readily take\nadvantage of feedback-prop without any retraining or fine-tuning. Our\nfeedback-prop inference procedure is general, simple, reliable, and works on\ndifferent challenging visual recognition tasks. We present two variants of\nfeedback-prop based on layer-wise and residual iterative updates. We experiment\nusing several multi-task models and show that feedback-prop is effective in all\nof them. Our results unveil a previously unreported but interesting dynamic\nproperty of deep CNNs. We also present an associated technical approach that\ntakes advantage of this property for inference under partial evidence in\ngeneral visual recognition tasks.","url_abs":"http://arxiv.org/abs/1710.08049v2","url_pdf":"http://arxiv.org/pdf/1710.08049v2.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":"feedback-prop-convolutional-neural-network","repo_url":"https://github.com/uvavision/feedbackprop","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1710.08049","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}