Code summarization aims at generating natural language abstraction for source code, and it can be of great help program comprehension software maintenance. The current code approaches have made progress with neural-network. However, most these methods focus on learning the semantic syntax snippets, ignoring dependency codes. In this paper, we propose a novel method based neural-network model using knowledge call between its related We extract dependencies from transform as token sequence...

In recent years, a lot of research has been conducted in the field object detection for aerial image, and many effective methods have emerged. This paper enhances existing Faster RCNN model to achieve better accuracy. We improve feature extraction using multi-scale fusion. Compared with traditional model, accuracy on mAP is improved by 1.06% applying proposed method. The visual effects numerical results verify improvement

With the rapid development of cloud computing, thousands servers and various applications are involved in data center. These changes result more complex flows center, which motivates need for faster, lower overhead, scalable large flow detection technology. This paper firstly shows shortcomings traditional technologies. Then it proposes a new method named EffiEye, efficiently realizes application-aware controller. EffiEye mainly replies on two different mechanisms: one is classification...

Social unrest events are common happenings in modern society which need to be proactively handled. An effective method is continuously assess the risk of upcoming social and predict likelihood these events. Our previous work built a hidden Markov model- (HMM-) based framework indicators associated with country instability, leaving two shortcomings can optimized: omitting event participants’ interaction implicitly learning state residence time. Inspired by this, we propose new prediction this...

Continuously assessing the risk of upcoming social unrest events and predict likelihood these are great importance. Thanks to era big data, people's understanding, experience, values, ideology mirrored in organization cyber space. In this paper, we propose an online prediction framework, using frequent subgraph patterns hidden semi-Markov models (HSMMs). The feature called BoEAG (Bag-of-Event-Association-subGraph) is constructed based on frequents mining bag word model. new framework...

Nowadays, research on social networks has attracted a large amount of attention from both academic and industrial societies. To understand the diffusion process guide viral marketing, it is importance to model then estimate influence seed user target user, which defined as in this paper. In famous models like independent cascade linear threshold model, tremendous computational costs are usually required estimating probability through simulation. paper, we adopt duplicate forwarding propose...

Image deblurring is a classic and important problem in industrial fields, such as aviation photo restoration, object recognition robotics, autonomous vehicles. Blurry images real‐world scenarios consist of mixed blurring types, natural motion owing to shaking the camera. Fast does not deblur entire image because it best option. Considering computational costs, also better have an alternative kernel different objects at high‐semantic level. To achieve restoration quality, beneficial combine...

<title>Abstract</title> Existing methods for video super-resolution (VSR) and frame interpolation (VFI) primarily concentrate on devising a general pipeline suitable open-domain videos. However, these approaches tend to overlook the inherent distinctions in animation data. Specifically, often features lines smooth areas that lack textures, thereby complicating estimation of inter-frame motions. Moreover, exaggerated expressions common introduce non-linear highly altered motion...

To quickly discover suspicious data in massive and facilitate the implementation of audit work, we need to use scientific experimental methods conduct in-depth analysis collected data. This experiment uses bisecting k-means clustering algorithm which is an unsupervised learning method realize on delivery 20 kinds drugs from one hospital 2018. During this experiment, three features drug quantity, retail price total are selected as indexes, that is, coordinate axes three-dimensional diagram,...

Inferring users' occupational categories on the basis of user-generated content becomes an important issue in user profiling and applications such as personalised recommendation systems with rapid explosion usage online social me-dia. Although previous work has demonstrated that language features extracted from media can effectively predict occupations, overcoming challenge time-consuming, expensive expert knowledge low prediction performance is fairly limited mostly based English platforms....

Software developers advance the project process by contributing and discussing on code platform. bot acts as an assistant to help deal with repetitive tasks. In this paper, we explore whether adoption of bots has impact developer sentiment projects progress. We collected issues, pull requests comments from GitHub popular projects. And found that human users had significantly reduced positive in bot-created issues PRs. The average merging time PRs is shorter than human-created ones. solving...

With the explosive growth of users’ online information, user profiling, inferring traits or interests, has attracted increasing attention due to its various applications in reliable personalized service and recommender system fields. Most existing studies regard profiling as a node classification task utilize graph-based methods exploit relations affiliated information heterogeneous graphs. However, they only consider single view pair-wise relationships between users, ignoring modeling...