The state-of-the-art acoustic modeling for Keyword Spotting (KWS) systems is mainly based on the hybrid model of Hidden Markov Model (HMM) and Neural Network (NN). However, it is challenging to efficiently train such a hybrid system, since the dependence of the intermediate phonetic representation. Motivated by the end-to-end speech recognition systems, we propose a Mandarin KWS system using the end-to-end method, which directly predict the posterior of phonetic units. The system is based on Connectionist Temporal Classifier (CTC) and Recurrent Neural Network (RNN). The main difference between our system and other CTC-based KWS system is the output alphabet and its corresponded keyword searching mechanism. We adopt Mandarin syllables as the output labels, rather than the phonemes or characters. Extensive experiments are conducted on the Mandarin Chinese speech dataset. Experimental results indicate that: (1) Compared with HMM-based KWS system, the end-to-end KWS system achieves a significant improvement, without any increase of computational cost. (2) Our syllable-based end-to-end KWS system obtains better performance than the state-of-the-art ones based on Chinese context independent (CI) phonemes or Chinese characters.

The discussion of collective agency involves the reduction problem of the concept of a collective. Individualism and Cartesian internalism have long restricted orthodox theories and made them face the tension between an irreducible concept of a collective and ontological reductionism. Heterodox theories as functionalism and interpretationism reinterpret the concept of agency and accept it as realized on the level of a collective. In order to adequately explain social phenomena that have relations as their essence, in this paper we propose a relational, holistic account of collective agency and argue that functionalism and interpretationism can be integrated into such an account.

To explore the relationship among attention concentration, interpersonal self-efficacy, Internet addiction and social network overuse among middle school students, a cluster sampling survey was carried out by using Yang’s Internet addiction scale, mobile social network overuse scale, ADHD self-examination scale and interpersonal self-efficacy scale among middle school students in four schools in three provinces and regions in China. The results showed that: firstly, Internet addiction was positively correlated with attention concentration and social network overuse; interpersonal self-efficacy was negatively correlated with internet addiction and attention concentration; social network overuse was positively correlated with attention concentration and interpersonal self-efficacy; secondly, when interpersonal self-efficacy, attention concentration and social network overuse predict Internet addiction, interpersonal self-efficacy significantly negatively predicts Internet addiction, attention concentration significantly positively predicts Internet addiction, and social network overuse significantly positively predicts Internet addiction; thirdly, attention concentration and interpersonal self-efficacy play a dual mediating role between social network overuse and Internet addiction.

Video-based moving vehicle detection and tracking are important parts of modern intelligent transportation system (ITS). They can provide valuable information such as vehicle velocity and trajectory for ITS. However, vehicle tracking at urban intersection is more challenging than that at highway, due to the complicated scenarios, such as the variety of vehicle moving direction, inter-vehicle clustering and occlusion. Many successful vehicle tracking systems developed for high way vehicle tracking based on the blob-tracking approach failed to provide acceptable performance at urban intersections when there are heavy vehicle occlusion or vehicles close to each other. This paper proposes a novel vehicle segmentation method for moving vehicle segmentation at urban intersection by seeking the spatial-temporal matching of feature points. Experimental results show that feature point may be taken as an important cue for moving vehicle segmentation and tracking under sophisticated traffic situation.

In order to effectively remove the noise existing in the transmission process of music signal, this paper discusses the music signal filtering algorithm from two different angles of time domain and frequency domain, and designs two equalization filters with adjustable parameters. And through the experimental simulation algorithm, the corresponding first-order and second-order low-frequency signal filtering cases are discussed. The effectiveness of the proposed method is demonstrated. At the same time, the method can be generalized to the case of high-order equalization filtering.