{"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/representation-similarity-analysis-for","title":"Representation Similarity Analysis for Efficient Task taxonomy & Transfer Learning","arxiv_id":"1904.11740","date":"2019-04-26","proceeding":"CVPR 2019 6","authors":["Kshitij Dwivedi","Gemma Roig"],"abstract":"Transfer learning is widely used in deep neural network models when there are\nfew labeled examples available. The common approach is to take a pre-trained\nnetwork in a similar task and finetune the model parameters. This is usually\ndone blindly without a pre-selection from a set of pre-trained models, or by\nfinetuning a set of models trained on different tasks and selecting the best\nperforming one by cross-validation. We address this problem by proposing an\napproach to assess the relationship between visual tasks and their\ntask-specific models. Our method uses Representation Similarity Analysis (RSA),\nwhich is commonly used to find a correlation between neuronal responses from\nbrain data and models. With RSA we obtain a similarity score among tasks by\ncomputing correlations between models trained on different tasks. Our method is\nefficient as it requires only pre-trained models, and a few images with no\nfurther training. We demonstrate the effectiveness and efficiency of our method\nfor generating task taxonomy on Taskonomy dataset. We next evaluate the\nrelationship of RSA with the transfer learning performance on Taskonomy tasks\nand a new task: Pascal VOC semantic segmentation. Our results reveal that\nmodels trained on tasks with higher similarity score show higher transfer\nlearning performance. Surprisingly, the best transfer learning result for\nPascal VOC semantic segmentation is not obtained from the pre-trained model on\nsemantic segmentation, probably due to the domain differences, and our method\nsuccessfully selects the high performing models.","url_abs":"http://arxiv.org/abs/1904.11740v1","url_pdf":"http://arxiv.org/pdf/1904.11740v1.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":"representation-similarity-analysis-for","repo_url":"https://github.com/kshitijd20/RSA-CVPR19-release","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"representation-similarity-analysis-for","repo_url":"https://github.com/cvai-repo/duality-diagram-similarity","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.11740","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}