Automatic Classification of Infant Vocalization Sequences with Convolutional Neural Networks
Speech Communication(2020)
摘要
•Small bottlenecks after the convolutional stage primarily raise CNN performance.•Global average pooling layers are most efficient for creating small bottlenecks.•Tuning the convolutional receptive field size is the second most important factor.•Vocalization classes are primarily confused according to their affective similarity.
更多查看译文
关键词
Infant,Vocalization,Convolutional neural network,Classification
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要