{"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/task-driven-convolutional-recurrent-models-of","title":"Task-Driven Convolutional Recurrent Models of the Visual System","arxiv_id":"1807.00053","date":"2018-06-20","proceeding":"NeurIPS 2018 12","authors":["Aran Nayebi","Daniel Bear","Jonas Kubilius","Kohitij Kar","Surya Ganguli","David Sussillo","James J. Dicarlo","Daniel L. K. Yamins"],"abstract":"Feed-forward convolutional neural networks (CNNs) are currently\nstate-of-the-art for object classification tasks such as ImageNet. Further,\nthey are quantitatively accurate models of temporally-averaged responses of\nneurons in the primate brain's visual system. However, biological visual\nsystems have two ubiquitous architectural features not shared with typical\nCNNs: local recurrence within cortical areas, and long-range feedback from\ndownstream areas to upstream areas. Here we explored the role of recurrence in\nimproving classification performance. We found that standard forms of\nrecurrence (vanilla RNNs and LSTMs) do not perform well within deep CNNs on the\nImageNet task. In contrast, novel cells that incorporated two structural\nfeatures, bypassing and gating, were able to boost task accuracy substantially.\nWe extended these design principles in an automated search over thousands of\nmodel architectures, which identified novel local recurrent cells and\nlong-range feedback connections useful for object recognition. Moreover, these\ntask-optimized ConvRNNs matched the dynamics of neural activity in the primate\nvisual system better than feedforward networks, suggesting a role for the\nbrain's recurrent connections in performing difficult visual behaviors.","url_abs":"http://arxiv.org/abs/1807.00053v2","url_pdf":"http://arxiv.org/pdf/1807.00053v2.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":"task-driven-convolutional-recurrent-models-of","repo_url":"https://github.com/neuroailab/tnn","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"object-recognition","task_name":"Object Recognition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1807.00053","atlas_url":"https://app.syntology.ai/?focus=1807.00053","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}