{"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/like-what-you-like-knowledge-distill-via","title":"Like What You Like: Knowledge Distill via Neuron Selectivity Transfer","arxiv_id":"1707.01219","date":"2017-07-05","proceeding":"ICLR 2019 5","authors":["Zehao Huang","Naiyan Wang"],"abstract":"Despite deep neural networks have demonstrated extraordinary power in various\napplications, their superior performances are at expense of high storage and\ncomputational costs. Consequently, the acceleration and compression of neural\nnetworks have attracted much attention recently. Knowledge Transfer (KT), which\naims at training a smaller student network by transferring knowledge from a\nlarger teacher model, is one of the popular solutions. In this paper, we\npropose a novel knowledge transfer method by treating it as a distribution\nmatching problem. Particularly, we match the distributions of neuron\nselectivity patterns between teacher and student networks. To achieve this\ngoal, we devise a new KT loss function by minimizing the Maximum Mean\nDiscrepancy (MMD) metric between these distributions. Combined with the\noriginal loss function, our method can significantly improve the performance of\nstudent networks. We validate the effectiveness of our method across several\ndatasets, and further combine it with other KT methods to explore the best\npossible results. Last but not least, we fine-tune the model to other tasks\nsuch as object detection. The results are also encouraging, which confirm the\ntransferability of the learned features.","url_abs":"http://arxiv.org/abs/1707.01219v2","url_pdf":"http://arxiv.org/pdf/1707.01219v2.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":"like-what-you-like-knowledge-distill-via","repo_url":"https://github.com/TuSimple/neuron-selectivity-transfer","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1707.01219","atlas_url":"https://app.syntology.ai/?focus=1707.01219","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}