{"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/e-swish-adjusting-activations-to-different","title":"E-swish: Adjusting Activations to Different Network Depths","arxiv_id":"1801.07145","date":"2018-01-22","proceeding":null,"authors":["Eric Alcaide"],"abstract":"Activation functions have a notorious impact on neural networks on both\ntraining and testing the models against the desired problem. Currently, the\nmost used activation function is the Rectified Linear Unit (ReLU). This paper\nintroduces a new and novel activation function, closely related with the new\nactivation $Swish = x * sigmoid(x)$ (Ramachandran et al., 2017) which\ngeneralizes it. We call the new activation $E-swish = \\beta x * sigmoid(x)$. We\nshow that E-swish outperforms many other well-known activations including both\nReLU and Swish. For example, using E-swish provided 1.5% and 4.6% accuracy\nimprovements on Cifar10 and Cifar100 respectively for the WRN 10-2 when\ncompared to ReLU and 0.35% and 0.6% respectively when compared to Swish. The\ncode to reproduce all our experiments can be found at\nhttps://github.com/EricAlcaide/E-swish","url_abs":"http://arxiv.org/abs/1801.07145v1","url_pdf":"http://arxiv.org/pdf/1801.07145v1.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":"e-swish-adjusting-activations-to-different","repo_url":"https://github.com/EricAlcaide/E-swish","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[],"methods":[{"method_slug":"e-swish","method_name":"E-swish"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"}],"datasets_introduced":[],"methods_introduced":[{"slug":"e-swish","name":"E-swish","full_name":"E-swish"}],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}