{"url":"/method/network-dissection","slug":"network-dissection","name":"Network Dissection","full_name":"Network Dissection","full_name_withheld":false,"description_markdown":"**Network Dissection** is an interpretability method for [CNNs](https://paperswithcode.com/methods/category/convolutional-neural-networks) that evaluates the alignment between individual hidden units and a set of visual semantic concepts. By identifying the best alignments, units are given human interpretable labels across a range of objects, parts, scenes, textures, materials, and colors. \r\n\r\nThe measurement of interpretability proceeds in three steps:\r\n\r\n- Identify a broad set of human-labeled visual concepts.\r\n- Gather the response of the hidden variables to known concepts.\r\n- Quantify alignment of hidden variable−concept pairs.","description_state":"present","introduced_year":null,"introduced_by":{"title":"Interpreting Deep Visual Representations via Network Dissection","paper":"/paper/interpreting-deep-visual-representations-via","first_author":"Bolei Zhou","n_authors":4,"url_abs":null,"archive_paper_url":"https://paperswithcode.com/paper/interpreting-deep-visual-representations-via"},"source":{"url":"http://arxiv.org/abs/1711.05611v2","title":"Interpreting Deep Visual Representations via Network Dissection","url_on_a_paper_host":true},"code_snippet_url":null,"code_snippet_url_on_a_code_host":false,"categories":[{"area":"General","area_id":"general","collection":"Interpretability","url":"/methods/category/interpretability","pwc_aliases":[]}],"n_papers_tagged":11,"archive_num_papers":11,"papers_newest_first":[{"paper":"/paper/2311-13594","title":"Labeling Neural Representations with Inverse Recognition","date":"2023-11-22","arxiv_id":"2311.13594","n_code_links":2,"syntology":{"ran":4,"of":4,"unverified":0,"pointer_only":4}},{"paper":"/paper/discover-making-vision-networks-interpretable","title":"DISCOVER: Making Vision Networks Interpretable via Competition and Dissection","date":"2023-09-21","arxiv_id":null,"n_code_links":1,"syntology":null},{"paper":null,"title":"On the Impact of Knowledge Distillation for Model Interpretability","date":"2023-05-25","arxiv_id":"2305.15734","n_code_links":0,"syntology":null},{"paper":"/paper/detection-accuracy-for-evaluating","title":"Detection Accuracy for Evaluating Compositional Explanations of Units","date":"2021-09-16","arxiv_id":"2109.07804","n_code_links":1,"syntology":null},{"paper":"/paper/interpreting-face-inference-models-using","title":"Interpreting Face Inference Models using Hierarchical Network Dissection","date":"2021-08-23","arxiv_id":"2108.10360","n_code_links":1,"syntology":null},{"paper":"/paper/gated-convolutional-networks-with-hybrid","title":"Gated Convolutional Networks with Hybrid Connectivity for Image Classification","date":"2019-08-26","arxiv_id":"1908.09699","n_code_links":1,"syntology":{"ran":4,"of":5,"unverified":1,"pointer_only":5}},{"paper":null,"title":"Interpreting Adversarial Examples by Activation Promotion and Suppression","date":"2019-04-03","arxiv_id":"1904.02057","n_code_links":0,"syntology":null},{"paper":null,"title":"On the Units of GANs (Extended Abstract)","date":"2019-01-29","arxiv_id":"1901.09887","n_code_links":0,"syntology":null},{"paper":"/paper/gan-dissection-visualizing-and-understanding","title":"GAN Dissection: Visualizing and Understanding Generative Adversarial Networks","date":"2018-11-26","arxiv_id":"1811.10597","n_code_links":8,"syntology":{"ran":4,"of":34,"unverified":30,"pointer_only":0}},{"paper":"/paper/how-convolutional-neural-network-see-the","title":"How convolutional neural network see the world - A survey of convolutional neural network visualization methods","date":"2018-04-30","arxiv_id":"1804.11191","n_code_links":1,"syntology":null},{"paper":"/paper/interpreting-deep-visual-representations-via","title":"Interpreting Deep Visual Representations via Network Dissection","date":"2017-11-15","arxiv_id":"1711.05611","n_code_links":3,"syntology":null}],"papers_shown":11,"tasks":[{"task":"/task/adversarial-defense","name":"Adversarial Defense","papers":1},{"task":"/task/adversarial-robustness","name":"Adversarial Robustness","papers":1},{"task":"/task/attribute","name":"Attribute","papers":1},{"task":"/task/classification-1","name":"Classification","papers":1},{"task":"/task/decision-making","name":"Decision Making","papers":1},{"task":"/task/classification","name":"General Classification","papers":1},{"task":"/task/image-classification","name":"Image Classification","papers":1},{"task":"/task/image-generation","name":"Image Generation","papers":1},{"task":"/task/image-retrieval","name":"Image Retrieval","papers":1},{"task":"/task/knowledge-distillation","name":"Knowledge Distillation","papers":1},{"task":"/task/object","name":"Object","papers":1},{"task":"/task/object-detection","name":"Object Detection","papers":1},{"task":"/task/retrieval","name":"Retrieval","papers":1},{"task":"/task/segmentation","name":"Segmentation","papers":1},{"task":"/task/image-classification","name":"image-classification","papers":1},{"task":"/task/object-detection-1","name":"object-detection","papers":1}],"tasks_shown":16,"n_tasks":16,"usage_by_year":[{"year":"2017","papers":1},{"year":"2018","papers":2},{"year":"2019","papers":3},{"year":"2021","papers":2},{"year":"2023","papers":3}],"row_source":"methods_table","archive":{"source":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","archive_url":"https://paperswithcode.com/method/network-dissection"},"syntology_read_at":"2026-09-24T18:15:14+00:00"}