基于换能器的语音识别的加速器感知训练

Suhaila M. Shakiah, R. Swaminathan, H. Nguyen, Raviteja Chinta, Tariq Afzal, Nathan Susanj, A. Mouchtaris, Grant P. Strimel, A. Rastrow
{"title":"基于换能器的语音识别的加速器感知训练","authors":"Suhaila M. Shakiah, R. Swaminathan, H. Nguyen, Raviteja Chinta, Tariq Afzal, Nathan Susanj, A. Mouchtaris, Grant P. Strimel, A. Rastrow","doi":"10.1109/SLT54892.2023.10022592","DOIUrl":null,"url":null,"abstract":"Machine learning model weights and activations are represented in full-precision during training. This leads to performance degradation in runtime when deployed on neural network accelerator (NNA) chips, which leverage highly parallelized fixed-point arithmetic to improve runtime memory and latency. In this work, we replicate the NNA operators during the training phase, accounting for the degradation due to low-precision inference on the NNA in back-propagation. Our proposed method efficiently emulates NNA operations, thus foregoing the need to transfer quantization error-prone data to the Central Processing Unit (CPU), ultimately reducing the user perceived latency (UPL). We apply our approach to Recurrent Neural Network-Transducer (RNN-T), an attractive architecture for on-device streaming speech recognition tasks. We train and evaluate models on 270K hours of English data and show a 5-7% improvement in engine latency while saving up to 10% relative degradation in WER.","PeriodicalId":352002,"journal":{"name":"2022 IEEE Spoken Language Technology Workshop (SLT)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Accelerator-Aware Training for Transducer-Based Speech Recognition\",\"authors\":\"Suhaila M. Shakiah, R. Swaminathan, H. Nguyen, Raviteja Chinta, Tariq Afzal, Nathan Susanj, A. Mouchtaris, Grant P. Strimel, A. Rastrow\",\"doi\":\"10.1109/SLT54892.2023.10022592\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Machine learning model weights and activations are represented in full-precision during training. This leads to performance degradation in runtime when deployed on neural network accelerator (NNA) chips, which leverage highly parallelized fixed-point arithmetic to improve runtime memory and latency. In this work, we replicate the NNA operators during the training phase, accounting for the degradation due to low-precision inference on the NNA in back-propagation. Our proposed method efficiently emulates NNA operations, thus foregoing the need to transfer quantization error-prone data to the Central Processing Unit (CPU), ultimately reducing the user perceived latency (UPL). We apply our approach to Recurrent Neural Network-Transducer (RNN-T), an attractive architecture for on-device streaming speech recognition tasks. We train and evaluate models on 270K hours of English data and show a 5-7% improvement in engine latency while saving up to 10% relative degradation in WER.\",\"PeriodicalId\":352002,\"journal\":{\"name\":\"2022 IEEE Spoken Language Technology Workshop (SLT)\",\"volume\":\"32 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2023-01-09\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2022 IEEE Spoken Language Technology Workshop (SLT)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/SLT54892.2023.10022592\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 IEEE Spoken Language Technology Workshop (SLT)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/SLT54892.2023.10022592","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

机器学习模型的权值和激活值在训练过程中以全精度表示。当部署在神经网络加速器(NNA)芯片上时,这会导致运行时性能下降,NNA芯片利用高度并行的定点算法来改善运行时内存和延迟。在这项工作中,我们在训练阶段复制了NNA算子,考虑了由于反向传播中对NNA的低精度推断而导致的退化。我们提出的方法有效地模拟了NNA操作,从而无需将量化容易出错的数据传输到中央处理单元(CPU),最终减少了用户感知延迟(UPL)。我们将我们的方法应用于递归神经网络传感器(RNN-T),这是一种用于设备上流语音识别任务的有吸引力的架构。我们在270K小时的英语数据上训练和评估了模型,结果显示发动机延迟改善了5-7%,同时减少了10%的相对下降。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Accelerator-Aware Training for Transducer-Based Speech Recognition
Machine learning model weights and activations are represented in full-precision during training. This leads to performance degradation in runtime when deployed on neural network accelerator (NNA) chips, which leverage highly parallelized fixed-point arithmetic to improve runtime memory and latency. In this work, we replicate the NNA operators during the training phase, accounting for the degradation due to low-precision inference on the NNA in back-propagation. Our proposed method efficiently emulates NNA operations, thus foregoing the need to transfer quantization error-prone data to the Central Processing Unit (CPU), ultimately reducing the user perceived latency (UPL). We apply our approach to Recurrent Neural Network-Transducer (RNN-T), an attractive architecture for on-device streaming speech recognition tasks. We train and evaluate models on 270K hours of English data and show a 5-7% improvement in engine latency while saving up to 10% relative degradation in WER.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信