Twitter sentiment analysis is an effective tool for various Twitter-based tasks. However, there still no neural-network-based research which takes both the tweet-text information and user-connection into account. To this end, we propose Attentional-graph Neural Network based Sentiment Analyzer (AGN-TSA), a analyzer on attentional-graph neural networks. AGN-TSA fuses through three-layered structure, includes word-embedding layer, user-embedding layer attentional graph network layer. For...

Graph-based approaches have been widely employed to facilitate in analyzing network flow connectivity behaviors, which aim understand the impacts and patterns of events. However, existing suffer from lack connectivity-behavior information loss event identification. In this paper, we propose graphs (NFCGs) capture behavior for modeling social behaviors entities. Given a set flows, edges NFCG are generated by connecting pairwise hosts who communicate with each other. To preserve more about...

Network worms spread widely over the global network within a short time, which are increasingly becoming one of most potential threats to security. However, performance traditional packet-oriented signature-based methods is questionable in face unknown worms, while anomaly-based approaches often exhibit high false positive rates. It common scenario that life cycle consists same four stages, target discovery phase and transferring have specific interactive patterns. To this end, we propose...

Identifying IDC (Internet Data Center) IP addresses and analyzing the connection relationship of could reflect network resource allocation layout which is helpful for optimization. Recent research mainly focuses on minimizing electricity consumption optimizing based traffic behavior analysis. However, lack network-wide information from operators has led to problems like management difficulties unbalanced IDC, are still unsolved today. In this paper, we propose a method identification...

The traffic-behavioral profiling of end-targets can provide the network administrators with user information both depictive and precise for better decision-making. Based on enumerated researches, this paper summarized basic conceptions end-targets, as well prevailing frameworks these techniques. Meanwhile, existing methods are carefully categorized, respective performances features contrasted, potential future researches introduced.

Predicting internet user demographics based on traffic behavior analysis can provide effective clues for the decision making of network administrators. Nonetheless, most existing researches overly rely hand-crafted features, and they also suffer from shallowness information mining limitation in prediction targets. This paper proposes Argus, a hierarchical neural solution to Internet through analysis. Argus is neural-network structure composed an autoencoder embedding fully-connected net...

Flow classification is of great significance for network management. Machine-learning-based flow widely used nowadays, but features which depict the non-Gaussian characteristics flows are still absent. In this paper, we propose Windowed Higher-order Statistical Analysis (WHOSA) machine-learning-based classification. our methodology, a modeled as three different time series: rate sequence, packet length sequence and inter-arrival sequence. For each both higher-order moments largest singular...

Lack of large available datasets fully annotated is a fundamental bottleneck in pulmonary nodule detection, especially when the corresponding computed tomography(CT) images obtained are device-dependent. We propose spatio-semantic attentive CycleGAN (SSA-CycleGAN) capable aligning modalities, as well distinguishing vs. non-nodule, which turn achieves effective data augmentation. Specifically, novel training loss function established, providing constraint for semantic preservation and local...

Much progress has been made in the deep neural network (DNN) based diagnosis of mass lesions breast ultrasound (BUS) images. However, non-mass lesion is less investigated because limited data. Based on insight that data sufficient and shares same knowledge structure with identifying malignancy a image, we propose novel transfer learning framework to enhance generalizability DNN model for BUS help BUS. Specifically, train shared combined With prior different marginal distributions input...

The warehousing and logistics industry is a basic, strategic leading that supports the development of national economy. Efforts must be made to improve intelligent level in Turnover Time. Warehousing power field are large scale wide scope. In this paper, we use exponential smoothing algorithm compress amounts data while eliminating extreme data. K-means DBSCAN algorithms used deal with factors related turnover time materials.

Much progress has been made in the deep neural network (DNN) based diagnosis of mass lesions breast ultrasound (BUS) images. However, non-mass lesion is less investigated because limited data. Based on insight that data sufficient and shares same knowledge structure with identifying malignancy a image, we propose novel transfer learning framework to enhance generalizability DNN model for BUS help BUS. Specifically, train shared combined With prior different marginal distributions input...