{"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/generalized-capsule-networks-with-trainable","title":"Generalized Capsule Networks with Trainable Routing Procedure","arxiv_id":"1808.08692","date":"2018-08-27","proceeding":"ICLR 2019 5","authors":["Zhenhua Chen","David Crandall"],"abstract":"CapsNet (Capsule Network) was first proposed by~\\citet{capsule} and later\nanother version of CapsNet was proposed by~\\citet{emrouting}. CapsNet has been\nproved effective in modeling spatial features with much fewer parameters.\nHowever, the routing procedures in both papers are not well incorporated into\nthe whole training process. The optimal number of routing procedure is misery\nwhich has to be found manually. To overcome this disadvantages of current\nrouting procedures in CapsNet, we embed the routing procedure into the\noptimization procedure with all other parameters in neural networks, namely,\nmake coupling coefficients in the routing procedure become completely\ntrainable. We call it Generalized CapsNet (G-CapsNet). We implement both\n\"full-connected\" version of G-CapsNet and \"convolutional\" version of G-CapsNet.\nG-CapsNet achieves a similar performance in the dataset MNIST as in the\noriginal papers. We also test two capsule packing method (cross feature maps or\nwith feature maps) from previous convolutional layers and see no evident\ndifference. Besides, we also explored possibility of stacking multiple capsule\nlayers. The code is shared on\n\\hyperlink{https://github.com/chenzhenhua986/CAFFE-CapsNet}{CAFFE-CapsNet}.","url_abs":"http://arxiv.org/abs/1808.08692v1","url_pdf":"http://arxiv.org/pdf/1808.08692v1.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":"generalized-capsule-networks-with-trainable","repo_url":"https://github.com/chenzhenhua986/CAFFE-CapsNet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[],"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}