Search

With the significant development of Internet, cyberspace has become one of the most important national sovereignties. To secure national cyberspace, it is essential to know the amount of cyberspace resources and understand the state and risk of the resources. Recently, researchers have developed enormous effort to devising new algorithms, strategies and tools to help detect and analyze cyberspace resources. Based on the relevant literature we surveyed, we defined the basic concepts of cyberspace resources and the framework of related techniques. Then we classified the main techniques and make a comparison between them, based on which we made an inductive review about the key technologies, the critical problems and possible solutions. We conclude with some research subjects that we hope will help improve the development of cyberspace resources scanning and analyzing, providing with more support for cyberspace security.

Given a set of facilities and a set of users, a reverse nearest neighbors (RNN) query retrieves every user $u$ for which the query facility $q$ is its closest facility. Since $q$ is the closest facility to $u$ , the user $u$ is said to be influenced by $q$ . In this paper, we propose a relaxed definition of influence where a user $u$ is said to be influenced by not only its closest facility but also every other facility that is almost as close to $u$ as its closest facility is. Based on this definition of influence, we propose reverse approximate nearest neighbors (RANN) queries. Formally, given a value $x>1$ , an RANN query $q$ returns every user $u$ for which $dist(u,q) \leq x\times NNDist(u)$ where $NNDist(u)$ denotes the distance between a user $u$ and its nearest facility, i.e., $q$ is an approximate nearest neighbor of $u$ . In this paper, we study both snapshot and continuous versions of RANN queries. In a snapshot RANN query, the underlying data sets do not change and the results of a query are to be computed only once. In the continuous version, the users continuously change their locations and the results of RANN queries are to be continuously monitored. Based on effective pruning techniques and several non-trivial observations, we propose efficient RANN query processing algorithms for both the snapshot and continuous RANN queries. We conduct extensive experiments on both real and synthetic data sets and demonstrate that our algorithm for both snapshot and continuous queries are significantly better than the competitors.

The visual background extractor (ViBe) has become one of the best motion object detection algorithms because of its good detection results and low memory requirements. However, the ViBe model cannot self-adjust the value range of the parameter that controls the number of samples chosen from the background template. In this paper, two models are proposed to help automatically change the parameter range in different environments. The blink energy model can detect dynamic backgrounds by increasing the range, while the object probability model can prevent corrosion of motion objects by decreasing the range. The experimental results show that our proposed method can both accurately recognize dynamic backgrounds and efficiently prevent object corrosion. In addition, our method shows better performance on benchmark datasets than several commonly used detection algorithms.

As a biometric trait, finger vein pattern-based technology is highly effective for personal identification with high security. In this paper, we presented the design of a personal identification system based on near infrared (NIR) finger vein image. In this paper, we introduced an observation model of finger vein imaging, upon which a self-adaptive illuminance control algorithm is proposed and integrated into image acquisition hardware. According to the distribution of pixel intensity of the acquired image, the proposed algorithm could automatically adjust the illuminance distribution of lighting: increase the illuminance of lighting, under which the thicker part of finger body is presented and decrease the illuminance of lighting, under which the thinner part of finger body is presented. With this adaptation, the whole finger body could be illuminated appropriately according to its thickness distribution, and the overexposure and underexposure are avoided effectively. An NIR finger vein image database containing 2040 images is established and published in this paper. In the image preprocessing stage, Gabor filters are used to enhance captured raw finger vein images. In our experiment, the identification performance of our system is evaluated using the recognition rate and the margin distribution. A sparse representation-based algorithm is used to calculate the recognition rate and provide data for margin analysis. The results prove the effectiveness of the proposed illuminance control algorithm and the whole system in finger vein-based personal identification.

We study the problem of structural graph clustering, a fundamental problem in managing and analyzing graph data. Given an undirected unweighted graph, structural graph clustering is to assign vertices to clusters, and to identify the sets of hub vertices and outlier vertices as well, such that vertices in the same cluster are densely connected to each other while vertices in different clusters are loosely connected. In this paper, we develop a new two-step paradigm for scalable structural graph clustering based on our three observations. Then, we present a $\mathsf {pSCAN}$ approach, within the paradigm, aiming to reduce the number of structural similarity computations, and propose optimization techniques to speed up checking whether two vertices are structure-similar. $\mathsf {pSCAN}$ outputs exactly the same clusters as the existing approaches $\mathsf {SCAN}$ and $\mathsf {SCAN\text{++}}$ , and we prove that $\mathsf {pSCAN}$ is worst-case optimal. Moreover, we propose efficient techniques for updating the clusters when the input graph dynamically changes, and we also extend our techniques to other similarity measures, e.g., Jaccard similarity. Performance studies on large real and synthetic graphs demonstrate the efficiency of our new approach and our dynamic cluster maintenance techniques. Noticeably, for the twitter graph with 1 billion edges, our approach takes 25 minutes while the state-of-the-art approach cannot finish even after 24 hours.

Semantic trajectories have become unprecedentedly available because of the rapidly growing popularities of location-sharing services. People’s lifestyles and Point-of-Interest demands are hidden in such data. Extracting people’s POI needs for different regions from semantic trajectories plays an important role in site selection, which can be widely used in city planning, facility location and other applications. However, most of existing works either use traditional trajectories which need to infer semantic with external information and lead to inaccuracy, or just focus on specific category. Semantic trajectory mining provides us a new way to address the challenges. Based on above motivation, we study the regional POI demand discovery problem using semantic trajectories. In this paper, we carefully analyze the features of semantic trajectory data and people’ mobility patterns. Then, we propose an effective POI demand modeling method. Furthermore, we propose two efficient algorithms to identify the regional POI demands. The proposed algorithms extract regional patterns and compute the regional POIs demand according to POI demand model. Finally, the ranked POIs demands for regions are obtained. We evaluate the proposed modeling method and algorithms in terms of efficiency and effectiveness on two real data sets. The results show that our proposed methods outperform the competitor for both efficiency and effectiveness.

© 2016. VLDB Endowment. Subgraph enumeration aims to find all the subgraphs of a large data graph that are isomorphic to a given pattern graph. As the subgraph isomorphism operation is computationally intensive, researchers have recently focused on solving this problem in distributed environments, such as MapReduce and Pregel. Among them, the state-of-the-art algorithm, Twin Twig Join, is proven to be instance optimal based on a left-deep join framework. However, it is still not scalable to large graphs because of the constraints in the left-deep join framework and that each decomposed component (join unit) must be a star. In this paper, we propose SEED - a scalable subgraph enumeration approach in the distributed environment. Compared to Twin Twig Join, SEED returns optimal solution in a generalized join framework without the constraints in Twin Twig Join. We use both star and clique as the join units, and design an effective distributed graph storage mechanism to support such an extension. We develop a comprehensive cost model, that estimates the number of matches of any given pattern graph by considering powerlaw degree distribution in the data graph. We then generalize the left-deep join framework and develop a dynamic-programming algorithm to compute an optimal bushy join plan. We also consider overlaps among the join units. Finally, we propose clique compression to further improve the algorithm by reducing the number of the intermediate results. Extensive performance studies are conducted on several real graphs, one containing billions of edges. The results demonstrate that our algorithm outperforms all other state-of-the-art algorithms by more than one order of magnitude.

The real estate search web systems such as Zillow, Anjuke, and Lianjia have become very popular in daily life. Generally, the comprehensive query results combined with transportation, health care, education, POIs, etc. are expected, but those surrounding information are rarely utilized in traditional query methods, which thereby restricts the results of the query. In this paper, we address the above limitations and provide a novel multi-view based query method, named KBHR. We investigate feature extraction method and introduce multi-view to represent comprehensive real estate data. The proposed method, KBHR, is based on BHR-tree which is a hybrid indexing structure and a kernel based similarity function developed to rank the query results of multi-view data. We construct experiments and evaluate KBHR on real-world data sets. The experimental results demonstrate the efficiency and effectiveness of our method.

Analyzing the structure of real-world networks has attracted much attention over years and cohesive subgraph models are commonly used to characterize the structure of a network. Recently, a model named k-Peak is proposed to address the issue failing to detect sparser regions if the network contains distinct regions of different densities in the cohesive subgraph models. However, k-Peak only considers the edge connection (i.e., degree) in the network and the loose structure restricts the effectiveness of the k-Peak. On the other hand, triangles are fundamental building blocks of a network and are widely used in the literature. Motivated by this, in this paper, we propose the k-TriPeak model based on the triangles and study the problem of k-TriPeak decomposition that computes the k-TriPeak for all possible k values to understand the structure of a network. Through investigating the drawbacks of the baseline algorithm following the idea of k-Peak decomposition, we devise a new efficient algorithm to perform the k-TriPeak decomposition. Our new algorithm adopts a top-down decomposition paradigm and integrates two novel upper bounds with which large unnecessary computation can be pruned. We conduct extensive experiments on several large real-world datasets and the experimental results demonstrate the efficiency and effectiveness of our proposed algorithm.

In this paper, we study the problem of set containment join. Given two collections $$\mathcal {R}$$ and $$\mathcal {S}$$ of records, the set containment join $$\mathcal {R} \bowtie _\subseteq \mathcal {S}$$ retrieves all record pairs $$\{(r,s)\} \in \mathcal {R}\times \mathcal {S}$$ such that $$r \subseteq s$$ . This problem has been extensively studied in the literature and has many important applications in commercial and scientific fields. Recent research focuses on the in-memory set containment join algorithms, and several techniques have been developed following intersection-oriented or union-oriented computing paradigms. Nevertheless, we observe that two computing paradigms have their limits due to the nature of the intersection and union operators. Particularly, intersection-oriented method relies on the intersection of the relevant inverted lists built on the elements of $$\mathcal {S}$$ . A nice property of the intersection-oriented method is that the join computation is verification free. However, the number of records explored during the join process may be large because there are multiple replicas for each record in $$\mathcal {S}$$ . On the other hand, the union-oriented method generates a signature for each record in $$\mathcal {R}$$ and the candidate pairs are obtained by the union of the inverted lists of the relevant signatures. The candidate size of the union-oriented method is usually small because each record contributes only one replica in the index. Unfortunately, union-oriented method needs to verify the candidate pairs, which may be cost expensive especially when the join result size is large. As a matter of fact, the state-of-the-art union-oriented solution is not competitive compared to the intersection-oriented ones. In this paper, we propose a new union-oriented method, namely TT-Join, which not only enhances the advantage of the previous union-oriented methods but also integrates the goodness of intersection-oriented methods by imposing a variant of prefix tree structure. We conduct extensive experiments on 20 real-life datasets and synthetic datasets by comparing our method with 7 existing methods. The experiment results demonstrate that TT-Join significantly outperforms the existing algorithms on most of the datasets and can achieve up to two orders of magnitude speedup. Furthermore, to support large scale of datasets, we extend our techniques to distributed systems on top of MapReduce framework. With the help of careful designed load-aware distribution mechanisms, our distributed join algorithm can achieve up to an order of magnitude speedup than the baselines methods.

© 2017 IEEE. In this paper, we study the problem of set containment join. Given two collections R and S of records, the set containment join R ⊆ S retrieves all record pairs {(r, s)} ∈ R × S such that r ⊆ s. This problem has been extensively studied in the literature and has many important applications in commercial and scientific fields. Recent research focuses on the in-memory set containment join algorithms, and several techniques have been developed following intersectionoriented or union-oriented computing paradigms. Nevertheless, we observe that two computing paradigms have their limits due to the nature of the intersection and union operators. Particularly, intersection-oriented method relies on the intersection of the relevant inverted lists built on the elements of S. A nice property of the intersection-oriented method is that the join computation is verification free. However, the number of records explored during the join process may be large because there are multiple replicas for each record in S. On the other hand, the unionoriented method generates a signature for each record in R and the candidate pairs are obtained by the union of the inverted lists of the relevant signatures. The candidate size of the union-oriented method is usually small because each record contributes only one replica in the index. Unfortunately, unionoriented method needs to verify the candidate pairs, which may be cost expensive especially when the join result size is large. As a matter of fact, the state-of-The-Art union-oriented solution is not competitive compared to the intersection-oriented ones. In this paper, we propose a new union-oriented method, namely TT-Join, which not only enhances the advantage of the previous unionoriented methods but also integrates the goodness of intersectionoriented methods by imposing a variant of prefix tree structure. We conduct extensive experiments on 20 real-life datasets by comparing our method with 7 existing methods. The experiment results demonstrate that TT-Join significantly outperforms the existing algorithms on most of the datasets, and can achieve up to two orders of magnitude speedup.

Pedestrian tracking plays an essential role in the domain of visual tracking. Much research in recent years has focused on how to obtain the accurate tracking results. However, few researchers have addressed the problem of the smoothness for the tracking trajectory. Most trajectory results are skipped and lack of smoothness, which does not comply with human vision habits. Also, some incorrect data has been recorded in the final trajectory due to the problem of partial occlusion. In this paper, we proposed a strategy to fix the present tracking trajectory. First, we detect and delete the anomaly tracking data. Second, we use a technique of supervised learning to do the linear regression for reconciling the tracking trajectory. The experiment manifests that the performance of the tracking trajectory is more accurate and stable than the original one after the implementation of our proposed strategy.

Today, in the construction of smart city, the development of self-driving technology plays the key role. The explosion of convolutional neural network (CNN) technology has made it possible to utilize end-to-end tasks with images. However, today’s CNN has deeper, more accurate characteristics. If we do not improve the calculation method to reduce the number of network parameters, this feature makes it very difficult for us to run neural network computing in small devices. In this paper, we further optimize the network computing methods based on MobileNets to reduce number of network parameters. At the same time, in the network structure, we add BatchNormalization and Swish activation function. We designed our own network in the end-to-end prediction for steering angle in the self-driving car task. From the final simulation results, our neural network’s storage space can be reduced and the execution speed of neural network can be improved while maintaining the accuracy of the neural network.

Order preserving estimation is an estimation method that can retain the original order of the population parameters of interest. It is an important tool in many applications such as data visualization. In this paper, we focus on the population mean as our primary estimation function, and propose effective query processing strategy that can preserve the estimated order to be correct with probabilistic guarantees. We define the cost function as the number of samples taken for all the groups, and our goal is to make the sample size as small as possible. We compare our methods with state-of-the-art near-optimal algorithm in the literature, and achieve up to \(80\,\%\) reduction in the total sample size.

Nowadays, users post a lot of their ordinary life records to online social sites. Rich social content covers discussion, interaction and communication activities etc. The social data provides insights into users’ interest, preference and communication aspects. An interesting problem is how to profile users’ occupation, i.e., professional categories. It has great values for users’ recommendation and personalized delivery services. However, it is very challenging, compared to gender or age prediction, due to the multiple categories and complex scenarios.

© Springer International Publishing AG 2016. With the popularity of geo-equipped devices and location based services, spatial-keyword publish/subscribe has emerged as a very important framework to disseminate real-time messages (e.g., geo-tagged e-coupon) to registered subscriptions (e.g., users interested in nearby promotions). While there are several work focusing on improving the efficiency of spatial-keyword publish/subscribe, their techniques fail to consider both the spatial and keyword distributions in a fine manner, thus lacking of scalability when coping with massive subscriptions. In this demonstration, we propose SKPS, a centralized in-memory spatial-keyword publish/ subscribe system, which exploits fully the spatial and keyword distributions of subscription workload during indexing construction, and employs an efficient message matching algorithm to disseminate each incoming message to relevant subscriptions in a real-time manner. We present a prototype of SKPS which provides users with a web-based interface to explore the message dissemination in publish/subscribe system.

In multi-camera video tracking, the tracking scene and tracking-target appearance can become complex, and current tracking methods use entirely different databases and evaluation criteria. Herein, for the first time to our knowledge, we present a universally applicable template library updating approach for multi-camera human tracking called multi-state self-learning template library updating (RS-TLU), which can be applied in different multi-camera tracking algorithms. In RS-TLU, self-learning divides tracking results into three states, namely steady state, gradually changing state, and suddenly changing state, by using the similarity of objects with historical templates and instantaneous templates because every state requires a different decision strategy. Subsequently, the tracking results for each state are judged and learned with motion and occlusion information. Finally, the correct template is chosen in the robust template library. We investigate the effectiveness of the proposed method using three databases and 42 test videos, and calculate the number of false positives, false matches, and missing tracking targets. Experimental results demonstrate that, in comparison with the state-of-the-art algorithms for 15 complex scenes, our RS-TLU approach effectively improves the number of correct target templates and reduces the number of similar templates and error templates in the template library.