With the booming of the mobile computing and web technology, virtual and intelligent mobile applications become increasingly popular, e.g. web computing and web-based information retrieval. However... Show More

With the booming of the mobile computing and web technology, virtual and intelligent mobile applications become increasingly popular, e.g. web computing and web-based information retrieval. However, under contemporary network conditions and web application environment, it remains a challenging problem to achieve a trade-off between algorithm complexity and hardware performance. In this article, a Talking Avatar architecture is presented based on third-party cloud services. First, the authors propose a cloud service based multi-level layered software framework, which consists of user interface layer, business logic layer and data layer. Second, human face synthesis, speech conversion and social sharing schemes are introduced to integrate third-party cloud services. Third, experimental results on Android platforms indicate that the proposed Talking Avatar can be served efficiently in terms of memory consumption as well as average response time. In addition, stronger functions are provided compared with existing methods.

Feature extraction is an important step in electroencephalogram (EEG) processing of motor imagery, and the feature extraction of EEG directly affects the final classification results. Through the... Show More

Feature extraction is an important step in electroencephalogram (EEG) processing of motor imagery, and the feature extraction of EEG directly affects the final classification results. Through the analysis of various feature extraction methods, this article finally selects Common Spatial Patterns (CSP) and wavelet packet analysis (WPA) to extract the feature and uses Support Vector Machine (SVM) to classify and compare these extracted features. For the EEG data provided by GRAZ University, the accuracy rate of feature extraction using CSP algorithm is 85.5%, and the accuracy rate of feature extraction using wavelet packet analysis is 92%. Then this paper analyzes the EEG data collected by Emotiv epoc+ system. The classification accuracy of wavelet packet extracted features can still be maintained at more than 80%, while the classification accuracy of CSP extracted feature is decreased obviously. Experimental results show that the method of wavelet packet analysis towards competition data and Emotiv epoc+ system data can both get a desirable outcome.

A method based on the leader-follower algorithm is proposed for transformation among the formations. The introduction of the greedy algorithm, behavior-based control and virtual structure help... Show More

A method based on the leader-follower algorithm is proposed for transformation among the formations. The introduction of the greedy algorithm, behavior-based control and virtual structure help realize the region division and the calculation of the distribution of the leaders and followers in target formation. Collision detection and collision avoidance are proposed to solve path conflicts with error free feedback and effectively maintain the stability of motion. The modeling of transformation is simulated by the shape from a line to a wedged, in which the formation is adjusted by the distance difference obtained by feedback. The experimental results show that it is feasible and effective to implement the formation conversion and formation control, and the system possess a better robustness and stability.

Chinese word segmentation is the basis of the Chinese natural language processing (NLP). With the development of the deep learning, various neural network models are applied to the Chinese word... Show More

Chinese word segmentation is the basis of the Chinese natural language processing (NLP). With the development of the deep learning, various neural network models are applied to the Chinese word segmentation. However, current neural network models have the characteristics of artificial feature extraction, nonstandard word-weight, inability to effectively use long-distance information and long training time of models in Chinese word segmentation. To solve a series of problems, this article presents a CNN-Bidirectional GRU-CRF neural network model (CNN Bidirectional GRU CRF Network, CBiGCN), which breaks through the limit of conventional method window, truly realizes end-to-end processing and applies to the neural network model by the five-Tag set method, bias-variable-weight greedy strategy and supplements by Goldstein-Armijo guidelines. Besides, this model, with simple structure, is easy to be operated. And it can automatically learn features, reduces large amounts of tasks on specific knowledge in the form of handcrafted features and data pre-processing, makes use of context information effectively. The authors set an experiment with two data corpuses for Chinese word segmentation to evaluate their system. The experiment verified their new model can obtain better Chinese word segmentation results and greatly reduce training time.

Inspired by the function, mechanism and efficiency of the visual nerve system of human beings, a revolutionary detection and reorganization method named YOLO is present to provide an accurate... Show More

Inspired by the function, mechanism and efficiency of the visual nerve system of human beings, a revolutionary detection and reorganization method named YOLO is present to provide an accurate, stable and fast arithmetic for a variety of targets, be it target detection for unmanned vehicle, car license recognition and optimization for surveillance. The traditional method for object detection is to reuse the classifier to implement detection, in contrast, the method named YOLOV2 process this problem by considering it in the mathematical area as a regression of spatially discrete bordered areas and relative class probability. However, as a cost of stable and fast response of this arithmetic, inaccurate detection maybe caused by YOLOV2 when the detected object is tiny (e.g., face recognition in surveillance). In this article, the authors provide a new method to further improve the performance of YOLOV2 by utilizing the accurate, stable and fast properties of YOLOV2 and editing the original code of YOLOV2 to eliminate the inaccuracy of tiny object detection, and implement this method on an embedded system.