{"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/neural-dataset-generality","title":"Neural Dataset Generality","arxiv_id":"1605.04369","date":"2016-05-14","proceeding":null,"authors":["Ragav Venkatesan","Vijetha Gattupalli","Baoxin Li"],"abstract":"Often the filters learned by Convolutional Neural Networks (CNNs) from\ndifferent datasets appear similar. This is prominent in the first few layers.\nThis similarity of filters is being exploited for the purposes of transfer\nlearning and some studies have been made to analyse such transferability of\nfeatures. This is also being used as an initialization technique for different\ntasks in the same dataset or for the same task in similar datasets.\nOff-the-shelf CNN features have capitalized on this idea to promote their\nnetworks as best transferable and most general and are used in a cavalier\nmanner in day-to-day computer vision tasks.\n  It is curious that while the filters learned by these CNNs are related to the\natomic structures of the images from which they are learnt, all datasets learn\nsimilar looking low-level filters. With the understanding that a dataset that\ncontains many such atomic structures learn general filters and are therefore\nuseful to initialize other networks with, we propose a way to analyse and\nquantify generality among datasets from their accuracies on transferred\nfilters. We applied this metric on several popular character recognition,\nnatural image and a medical image dataset, and arrived at some interesting\nconclusions. On further experimentation we also discovered that particular\nclasses in a dataset themselves are more general than others.","url_abs":"http://arxiv.org/abs/1605.04369v1","url_pdf":"http://arxiv.org/pdf/1605.04369v1.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":"neural-dataset-generality","repo_url":"https://github.com/ragavvenkatesan/generality","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"transfer-learning","task_name":"Transfer Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}