School of Information Science and Engineering, Henan University of Technology, Zhengzhou, China.
Key Laboratory of Underwater Acoustic signal Processing of Ministry of Education, Southeast University, Nanjing, China.
PLoS One. 2019 Oct 8;14(10):e0223361. doi: 10.1371/journal.pone.0223361. eCollection 2019.
Existing algorithms of speech-based deception detection are severely restricted by the lack of sufficient number of labelled data. However, a large amount of easily available unlabelled data has not been utilized in reality. To solve this problem, this paper proposes a semi-supervised additive noise autoencoder model for deception detection. This model updates and optimizes the semi-supervised autoencoder and it consists of two layers of encoder and decoder, and a classifier. Firstly, it changes the activation function of the hidden layer in network according to the characteristics of the deception speech. Secondly, in order to prevent over-fitting during training, the specific ratio dropout is added at each layer cautiously. Finally, we directly connected the supervised classification task in the output of encoder to make the network more concise and efficient. Using the feature set specified by the INTERSPEECH 2009 Emotion Challenge, the experimental results on Columbia-SRI-Colorado (CSC) corpus and our own deception corpus show that the proposed model can achieve more advanced performance than other alternative methods with only a small amount of labelled data.
现有的基于语音的欺骗检测算法受到缺乏足够数量的标记数据的严重限制。然而，大量易于获得的未标记数据在现实中尚未得到利用。为了解决这个问题，本文提出了一种用于欺骗检测的半监督加性噪声自动编码器模型。该模型对半监督自动编码器进行了更新和优化，它由两层编码器和解码器以及一个分类器组成。首先，根据欺骗性语音的特点，它改变了网络中隐藏层的激活函数。其次，为了防止训练过程中的过拟合，在每个层谨慎地添加了特定比例的 dropout。最后，我们直接在编码器的输出端连接监督分类任务，使网络更加简洁高效。使用 INTERSPEECH 2009 情感挑战指定的特征集，在哥伦比亚- SRI -科罗拉多州（CSC）语料库和我们自己的欺骗语料库上的实验结果表明，与其他替代方法相比，该模型仅使用少量标记数据就能实现更先进的性能。
