{"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/shifting-mean-activation-towards-zero-with","title":"Shifting Mean Activation Towards Zero with Bipolar Activation Functions","arxiv_id":"1709.04054","date":"2017-09-12","proceeding":"ICLR 2018 1","authors":["Lars Eidnes","Arild Nøkland"],"abstract":"We propose a simple extension to the ReLU-family of activation functions that\nallows them to shift the mean activation across a layer towards zero. Combined\nwith proper weight initialization, this alleviates the need for normalization\nlayers. We explore the training of deep vanilla recurrent neural networks\n(RNNs) with up to 144 layers, and show that bipolar activation functions help\nlearning in this setting. On the Penn Treebank and Text8 language modeling\ntasks we obtain competitive results, improving on the best reported results for\nnon-gated networks. In experiments with convolutional neural networks without\nbatch normalization, we find that bipolar activations produce a faster drop in\ntraining error, and results in a lower test error on the CIFAR-10\nclassification task.","url_abs":"http://arxiv.org/abs/1709.04054v3","url_pdf":"http://arxiv.org/pdf/1709.04054v3.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":"shifting-mean-activation-towards-zero-with","repo_url":"https://github.com/larspars/word-rnn","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}