{"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/interpreting-deep-visual-representations-via","title":"Interpreting Deep Visual Representations via Network Dissection","arxiv_id":"1711.05611","date":"2017-11-15","proceeding":null,"authors":["Bolei Zhou","David Bau","Aude Oliva","Antonio Torralba"],"abstract":"The success of recent deep convolutional neural networks (CNNs) depends on\nlearning hidden representations that can summarize the important factors of\nvariation behind the data. However, CNNs often criticized as being black boxes\nthat lack interpretability, since they have millions of unexplained model\nparameters. In this work, we describe Network Dissection, a method that\ninterprets networks by providing labels for the units of their deep visual\nrepresentations. The proposed method quantifies the interpretability of CNN\nrepresentations by evaluating the alignment between individual hidden units and\na set of visual semantic concepts. By identifying the best alignments, units\nare given human interpretable labels across a range of objects, parts, scenes,\ntextures, materials, and colors. The method reveals that deep representations\nare more transparent and interpretable than expected: we find that\nrepresentations are significantly more interpretable than they would be under a\nrandom equivalently powerful basis. We apply the method to interpret and\ncompare the latent representations of various network architectures trained to\nsolve different supervised and self-supervised training tasks. We then examine\nfactors affecting the network interpretability such as the number of the\ntraining iterations, regularizations, different initializations, and the\nnetwork depth and width. Finally we show that the interpreted units can be used\nto provide explicit explanations of a prediction given by a CNN for an image.\nOur results highlight that interpretability is an important property of deep\nneural networks that provides new insights into their hierarchical structure.","url_abs":"http://arxiv.org/abs/1711.05611v2","url_pdf":"http://arxiv.org/pdf/1711.05611v2.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":"interpreting-deep-visual-representations-via","repo_url":"https://github.com/csailvision/netdissect-lite","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"interpreting-deep-visual-representations-via","repo_url":"https://github.com/winycg/HCGNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"interpreting-deep-visual-representations-via","repo_url":"https://github.com/krlgroup/clustered-compositional-explanations","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[],"methods":[{"method_slug":"interpretability","method_name":"Interpretability"},{"method_slug":"network-dissection","method_name":"Network Dissection"}],"datasets_introduced":[],"methods_introduced":[{"slug":"network-dissection","name":"Network Dissection","full_name":"Network Dissection"}],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1711.05611","atlas_url":"https://app.syntology.ai/?focus=1711.05611","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}