{"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/expert-identification-of-visual-primitives","title":"Expert identification of visual primitives used by CNNs during mammogram classification","arxiv_id":"1803.04858","date":"2018-03-13","proceeding":null,"authors":["Jimmy Wu","Diondra Peck","Scott Hsieh","Vandana Dialani","Constance D. Lehman","Bolei Zhou","Vasilis Syrgkanis","Lester Mackey","Genevieve Patterson"],"abstract":"This work interprets the internal representations of deep neural networks\ntrained for classification of diseased tissue in 2D mammograms. We propose an\nexpert-in-the-loop interpretation method to label the behavior of internal\nunits in convolutional neural networks (CNNs). Expert radiologists identify\nthat the visual patterns detected by the units are correlated with meaningful\nmedical phenomena such as mass tissue and calcificated vessels. We demonstrate\nthat several trained CNN models are able to produce explanatory descriptions to\nsupport the final classification decisions. We view this as an important first\nstep toward interpreting the internal representations of medical classification\nCNNs and explaining their predictions.","url_abs":"http://arxiv.org/abs/1803.04858v1","url_pdf":"http://arxiv.org/pdf/1803.04858v1.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":"expert-identification-of-visual-primitives","repo_url":"https://github.com/jimmyyhwu/ddsm-visual-primitives","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}