{"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/improving-speech-recognition-by-revising","title":"Improving speech recognition by revising gated recurrent units","arxiv_id":"1710.00641","date":"2017-09-29","proceeding":null,"authors":["Mirco Ravanelli","Philemon Brakel","Maurizio Omologo","Yoshua Bengio"],"abstract":"Speech recognition is largely taking advantage of deep learning, showing that\nsubstantial benefits can be obtained by modern Recurrent Neural Networks\n(RNNs). The most popular RNNs are Long Short-Term Memory (LSTMs), which\ntypically reach state-of-the-art performance in many tasks thanks to their\nability to learn long-term dependencies and robustness to vanishing gradients.\nNevertheless, LSTMs have a rather complex design with three multiplicative\ngates, that might impair their efficient implementation. An attempt to simplify\nLSTMs has recently led to Gated Recurrent Units (GRUs), which are based on just\ntwo multiplicative gates.\n  This paper builds on these efforts by further revising GRUs and proposing a\nsimplified architecture potentially more suitable for speech recognition. The\ncontribution of this work is two-fold. First, we suggest to remove the reset\ngate in the GRU design, resulting in a more efficient single-gate architecture.\nSecond, we propose to replace tanh with ReLU activations in the state update\nequations. Results show that, in our implementation, the revised architecture\nreduces the per-epoch training time with more than 30% and consistently\nimproves recognition performance across different tasks, input features, and\nnoisy conditions when compared to a standard GRU.","url_abs":"http://arxiv.org/abs/1710.00641v1","url_pdf":"http://arxiv.org/pdf/1710.00641v1.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":"improving-speech-recognition-by-revising","repo_url":"https://github.com/mravanelli/theano-kaldi-rnn","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"speech-recognition","task_name":"Speech Recognition"},{"task_slug":"speech-recognition-1","task_name":"speech-recognition"}],"methods":[{"method_slug":"gru","method_name":"GRU"},{"method_slug":"relu","method_name":"ReLU"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}