{"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/choose-your-neuron-incorporating-domain","title":"Choose Your Neuron: Incorporating Domain Knowledge through Neuron-Importance","arxiv_id":"1808.02861","date":"2018-08-08","proceeding":"ECCV 2018 9","authors":["Ramprasaath R. Selvaraju","Prithvijit Chattopadhyay","Mohamed Elhoseiny","Tilak Sharma","Dhruv Batra","Devi Parikh","Stefan Lee"],"abstract":"Individual neurons in convolutional neural networks supervised for\nimage-level classification tasks have been shown to implicitly learn\nsemantically meaningful concepts ranging from simple textures and shapes to\nwhole or partial objects - forming a \"dictionary\" of concepts acquired through\nthe learning process. In this work we introduce a simple, efficient zero-shot\nlearning approach based on this observation. Our approach, which we call Neuron\nImportance-AwareWeight Transfer (NIWT), learns to map domain knowledge about\nnovel \"unseen\" classes onto this dictionary of learned concepts and then\noptimizes for network parameters that can effectively combine these concepts -\nessentially learning classifiers by discovering and composing learned semantic\nconcepts in deep networks. Our approach shows improvements over previous\napproaches on the CUBirds and AWA2 generalized zero-shot learning benchmarks.\nWe demonstrate our approach on a diverse set of semantic inputs as external\ndomain knowledge including attributes and natural language captions. Moreover\nby learning inverse mappings, NIWT can provide visual and textual explanations\nfor the predictions made by the newly learned classifiers and provide neuron\nnames. Our code is available at\nhttps://github.com/ramprs/neuron-importance-zsl.","url_abs":"http://arxiv.org/abs/1808.02861v1","url_pdf":"http://arxiv.org/pdf/1808.02861v1.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":"choose-your-neuron-incorporating-domain","repo_url":"https://github.com/ramprs/neuron-importance-zsl","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"generalized-zero-shot-learning","task_name":"Generalized Zero-Shot Learning"},{"task_slug":"zero-shot-learning","task_name":"Zero-Shot Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1808.02861","atlas_url":"https://app.syntology.ai/?focus=1808.02861","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}