Maxout neurons for deep convolutional and LSTM neural networks in speech recognition
详细信息 查看全文
- 作者：Meng Cai ; cai-m10@mails.tsinghua.edu.cn" class="auth_mail" title="E-mail the corresponding author ; caimeng06@foxmail.com" class="auth_mail" title="E-mail the corresponding author ; Jia Liu liuj@tsinghua.edu.cn" class="auth_mail" title="E-mail the corresponding author
- 关键词：Maxout neuron ; Convolutional neural network ; Long short-term memory ; Acoustic modeling ; Speech recognition
- 刊名：Speech Communication
- 出版年：2016
- 出版时间：March 2016
- 年：2016
- 卷：77
- 期：Complete
- 页码：53-64
- 全文大小：947 K
文摘
- •
We combine maxout neurons with convolutional and LSTM structures for DNNs.
- •
The optimal network structures and training strategies are explored for the models.
- •
Experiments are carried out for 6 languages on the IARPA Babel data sets.
- •
State-of-the-art results are achieved after system combination.