Aspect-based sentiment classification (ABSC) aims to predict sentiment polarities of different aspects within sentences or documents. Many previous studies have been conducted to solve this problem, but previous works fail to notice the correlation between the aspect’s sentiment polarity and the local context. In this paper, a Local Context Focus (LCF) mechanism is proposed for aspect-based sentiment classification based on Multi-head Self-Attention (MHSA). This mechanism is called LCF design, and utilizes the Context features Dynamic Mask (CDM) and Context Features Dynamic Weighted (CDW) layers to pay more attention to the local context words. Moreover, a BERT-shared layer is adopted to LCF design to capture internal long-term dependencies of local context and global context. Experiments are conducted on three common ABSC datasets: the laptop and restaurant datasets of SemEval-2014 and the ACL twitter dataset. Experimental results demonstrate that the LCF baseline model achieves considerable performance. In addition, we conduct ablation experiments to prove the significance and effectiveness of LCF design. Especially, by incorporating with BERT-shared layer, the LCF-BERT model refreshes state-of-the-art performance on all three benchmark datasets.

No abstract

No abstract

In the study of power control and allocation based on pricing, the utility of secondary users is usually studied from the perspective of the signal to noise ratio. The study of secondary user utility from the perspective of communication demand can not only promote the secondary users to meet the maximum communication needs, but also to maximize the utilization of spectrum resources, however, research in this area is lacking, so from the viewpoint of meeting the demand of network communication, this paper designs a two stage model to solve spectrum leasing and allocation problem in cognitive radio sensor networks (CRSNs). In the first stage, the secondary base station collects the secondary network communication requirements, and rents spectrum resources from several primary base stations using the Bertrand game to model the transaction behavior of the primary base station and secondary base station. The second stage, the subcarriers and power allocation problem of secondary base stations is defined as a nonlinear programming problem to be solved based on Nash bargaining. The simulation results show that the proposed model can satisfy the communication requirements of each user in a fair and efficient way compared to other spectrum sharing schemes.

No abstract

With the development of Internet and computer science, computer network is changing people's lives. Meanwhile, Network traffic prediction model itself becomes more and more complex. It is an important research direction to quickly and accurately detect the intrusions or attacks. The performance efficiency of a network intrusion detection system is dominated by pattern matching algorithm. However, In view of the complex non-linear and chaotic network traffic, and combined with its time-series properties, this paper propose a network traffic prediction model based on the auto-regressive moving average (ARMA). This model adopts the third party inspection systems which aim to save the network resources and predict the network traffic in high efficiency. By taking the data flow measurements in 16 channel analyzer to initialize the model, the simulation results show that the proposed model can effectively detect network intrusions and attacks, which improves the performance of the entire network and prolongs the network lifetime.

This paper describes an effective on-line polymer characterization technique by using small-angle light-scattering (SALS) image processing software and wavelet analysis. The phenomenon of small-angle light scattering has been applied to give information about transparent structures on morphology. Real-time visualization of various scattered light image and light intensity matrices is performed by the optical image real-time processing software for SALS. The software can measure the signal intensity of light scattering images, draw the frequency-intensity curves and the amplitude-intensity curves to indicate the variation of the intensity of scattered light in different processing conditions, and estimate the parameters. The current study utilizes a one-dimensional wavelet to delete noise from the original SALS signal and estimate the variation trend of maximum intensity area of the scattered light. So, the system brought the qualitative analysis of the structural information of transparent film success.

No abstract