Emerging Intention Mining Inspired by Semantic Reconstruction

Emerging intention mining is the process of identifying the inputs that differ in some respect from the training samples. With the development of artificial intelligence, more and more attention has been paid to data mining. Therefore, methods which can efficiently identify the emerging intentions are needed. However, due to the lack of data for emerging intention, training an end-to-end deep network is a cumbersome task. In this paper, we propose an end-to-end architecture for emerging intention mining which inspired by the success of generative adversarial networks and conditional variational autoencoder (CVAE). Our architecture tries understand the underlying concept of the inputs and then reconstruct them through semantic information. There are two deep networks by Convolutional Neural Networks (CNN). One network works as the semantic information extraction and the other works as the novelty detector. The first network supports the second by enhancing the inlier samples and distorting the outliers. The proposed framework applies to different datasets. The results illustrate that our proposed method learns the target class effectively and is superior to the baseline and state-of-the-art methods.