International Journal of Pattern Recognition and Artificial Intelligence

For the impact of the bitrate change of video streaming services according to the available bandwidth on user satisfaction, in this paper, we propose a spatial and temporal feature-based reduced reference (RR) quality assessment for rate-varying videos in wireless networks called STRQAW. First, simulating the orientation selectivity mechanism of the human visual system (HVS), the histogram of the orientation selectivity-based visual pattern in each frame is extracted as the spatial feature. The histogram similarity between the rate-varying video and the original video is computed as the spatial metric. Second, we extract the temporal variation of the DCT coefficients of the consecutive frame differences as the temporal feature. The temporal variation similarity between the rate-varying video and the original video is calculated as the temporal metric. Finally, we take into account the recency effect and assess the overall quality by combining the temporal and spatial metric. The experimental results using the Laboratory for Image and Video Engineering (LIVE) mobile video quality assessment (VQA) database show that STRQAW is consistent with the subjective assessment results, which means it reflects human subjective feelings well and it provides an evaluation for adjusting compression-coding rates in real time. STRQAW can be used to guide video application providers and network operators working towards satisfying end-user experiences.

In recent years, how to use renewable energy to reduce the energy cost of internet data center (IDC) has been an urgent problem to be solved. More and more solutions are beginning to consider machine learning, but many of the existing methods need to take advantage of some future information, which is difficult to obtain in the actual operation process. In this paper, we focus on reducing the energy cost of IDC by controlling the energy flow of renewable energy without any future information. we propose an efficient energy dynamic control algorithm based on the theory of reinforcement learning, which approximates the optimal solution by learning the feedback of historical control decisions. For the purpose of avoiding overestimation, improving the convergence ability of the algorithm, we use the double $Q$-method to further optimize. The extensive experimental results show that our algorithm can on average save the energy cost by 18.3% and reduce the rate of grid intervention by 26.2% compared with other algorithms, and thus has good application prospects.

Large-scale tasks processing based on cloud computing has become crucial to big data analysis and disposal in recent years. Most previous work, generally, utilize the conventional methods and architectures for general scale tasks to achieve tons of tasks disposing, which is limited by the issues of computing capability, data transmission, etc. Based on this argument, a fat-tree structure-based approach called LTDR (Large-scale Tasks processing using Deep network model and Reinforcement learning) has been proposed in this work. Aiming at exploring the optimal task allocation scheme, a virtual network mapping algorithm based on deep convolutional neural network and $Q$-learning is presented herein. After feature extraction, we design and implement a policy network to make node mapping decisions. The link mapping scheme can be attained by the designed distributed value-function based reinforcement learning model. Eventually, tasks are allocated onto proper physical nodes and processed efficiently. Experimental results show that LTDR can significantly improve the utilization of physical resources and long-term revenue while satisfying task requirements in big data.

In the natural language environment, accurately inferring the meaning of a token according to its context is crucial to understanding a sophisticated expression. However, this is not easy for a machine. The traditional language models used to train distributed word vectors are often restricted by single-sense embedding. In this paper, we develop a model called MSCvec (Multi-sense Soft Clustering Vector) for word sense disambiguation of polysemy in context. We extract the features of individual words by the co-occurrence PPMI (Positive Pointwise Mutual Information) matrix, and decompose the matrix by NMF (Nonnegative Matrix Factorization) into low-rank representations of target words, which are used as the input of an unsupervised sparse soft clustering method called Sparse Fuzzy C-means (SFCM). We use SFCM to determine the global semantic space of words, and partition the subspaces of multiple senses of a polysemous word. We relabel candidate words by the negative average log likelihood, and train multi-sense embedding with extensional vocabulary by the fastText model. Compared with the traditional static embeddings, the result shows that NMF and SFCM design can improve the performance in word similarity and relatedness tasks as well as in text classification tasks of different types of text. Accurate semantic representation of MSCvec would be necessary to produce outstanding results.

In heterogeneous wireless sensor networks, the data collection method based on compressed sensing technology is susceptible to packet loss and noise, which leads to a decrease in data reconstruction accuracy in unreliable links. Combining compressed sensing and matrix completion, we propose a clustering optimization algorithm based on structured noise matrix completion, in which the cluster head transmits the compressed sampling data and compression strategy to the base station. The algorithm we proposed can reduce the energy consumption of the node in the process of data collection, redundant data and transmission delay. The rank-1 matrix completion algorithm constructs an extremely sparse observation matrix, which is adopted by the sink node to complete the reconstruction of the whole network data. Simulation experiments show that the proposed algorithm reduces network transmission data, balances node energy consumption, improves data transmission efficiency and reconstruction accuracy, and extends the network life cycle.

Substation patrol robots (SIR) play an increasingly important role in ensuring the safe operation of substations. The robust and precise position estimating of the instruments to be inspected on the images are a prerequisite for accurately detecting the target states or obtaining the target readings under all-weather environment. In order to achieve high location accuracy of instrument, this study proposed an improved kernelized correlation filter (KCF) algorithm for achieving robust instrument location on images for SPR. Firstly, multiple templates are selected for training KCF classifier parameters. Then, reliable SURF matching-point determination method is designed, and the regions including reliable matching points are selected as the candidate regions, so that the searching range is narrowed. Finally, for KCF response surface of each candidate region, Single-Peak Constraint (SPC) is designed for locating target and reducing mismatching rate. Furthermore, experiments are performed for validating the effectiveness of the proposed algorithm, in which four instruments mainly including lightning arrester monitor and transformer thermometer are selected. The experimental results show that the proposed method has higher accuracy of target localization than traditional SURF-based position estimating method.

Single sample face recognition (SSFR) is a challenging research problem in which only one face image per person is available for training. Moreover, the face image may have different pose, expression, illumination, occlusion etc. rendering this problem more complex. Several methods have been suggested by various researchers in literature to solve SSFR. Here, we provide a comprehensive review of the methods proposed in the last decade for solving SSFR problem and introduce a novel taxonomy for the same. We divide SSFR methods broadly into five categories viz. (i) feature based, (ii) virtual sample generation based, (iii) generic database based, (iv) Hybrid and (v) other methods. We have also briefly reviewed the face databases used for evaluating single sample face recognition methods. Furthermore, the performance of the methods has been analyzed in terms of classification accuracy as given in literature. At last, we also suggest some future direction to the researchers and practitioners working in this fascinating research area.

With the development of computers and network technologies, network security has gradually become a global problem. Network security defenses need to be carried out not only on the Internet, but also on other communication media, such as electromagnetic signals. Existing electromagnetic signal communication is easily intercepted or infiltrated. In order to effectively detect the abnormal electromagnetic signal to find out the specific location, then classify it, it is necessary to study the way of communication. The existing electromagnetic signal detection accuracy is low and cannot be located. Considering the characteristics of different power sources in different locations, combined with spark streaming technology and machine learning classification technology, a joint platform for electromagnetic signal anomaly detection based on big data analysis is proposed. The electromagnetic signal is abnormally detected by feature comparison and small signal analysis, and the position and number between the signal sources are determined by three-point positioning and signal attenuation. The experimental results show that the method can detect abnormal electromagnetic signals and classify abnormal electromagnetic signals well, the accuracy rate can reach 95%, and the positioning accuracy can reach 89%.

To prevent any negative electromagnetic influence of high-density integrated circuits, an insulation package needs to be specially designed to shield it. Aiming at the low efficiency and material waste in traditional packaging methods, a printed circuit board (PCB) selective packaging system based on a multi-pattern solder joint simultaneous segmentation algorithm and three-dimensional printing technology is introduced in this paper. Firstly, the structure of PCB selective packaging system is designed. Secondly, to solve the existing problems, such as multi-pattern solder joints which are located densely in small welding areas and are hard to be extracted in the small-area integrated circuit board, a multi-pattern solder joint simultaneous segmentation algorithm is developed based on (geometrical) neighborhood features to extract and locate the optimal PCB solder joint areas. Finally, tests using three actual PCB are carried out to compare the proposed method with traditional multi-threshold solder joint extraction methods. Test results indicate that the proposed algorithm is simple and effective. Diverse solder joints can be optimally located and simultaneously extracted from the collected PCB image, which greatly improves the filling rate of the solder joint areas and filters out false pixels. Thus, this method provides a reliable location-finding tool to help place solder points in PCB selective packaging systems.

The rate-requirement of device-to-device (D2D) users is associated with the context information of velocity and data size of users to some extent. In this study, an efficient context-aware resource allocation scheme based on rate requirement (RARR) is proposed. This scheme consists of two allocation phases. In the rate-ensuring resource allocation phase, D2D pairs are allocated a certain amount of spectrum resource according to their rate requirement. In the allocation, the interference restricted area is limited to exclude cellular users that bring a negative capacity gain to the communication system. In the residual resource reallocation phase, surplus resources are assigned to D2D pairs according to the system fairness. Simulation results indicate that the proposed RARR scheme efficiently leads to superior performance in terms of system throughput and fairness and exhibits low complexity relative to traditional resource allocation.

Airfoil optimization algorithm is studied and a hybrid PSO and GA method is proposed in this paper. After function test, it shows that algorithm is well in convergence performance, fast speed, and optimization capability. Then, the airfoil parametric expression theory is analyzed. A new airfoil is obtained after combining CFD and PSO-GA optimization. The aerodynamic of new airfoil is compared with the airfoil optimized by GA-PSO and basic airfoil NACA0018. The results indicate that new airfoil is better than the other two airfoils in lift coefficient, lift-drag ratio, and surface pressure. At last, wing-sail of new airfoil and NACA0018 wing-sail are designed and manufactured. Both of them are applied in land-yacht robot linear motion and steering motion experiment. For the linear motion, in the situation of wind speed being 15m/s and angle of attack being 5, running speed of robot with optimized new wing-sail is 1.853m/s. In steering motion, trajectory with new wing-sail is closer to the real situation and it gets more thrust. The experiments data verify that the simulation results are correct and PSO-GA algorithm is effective.

This paper studies a searching problem in an unknown street. A simple polygon $P$ with two distinguished vertices, $s$ and $t$, is called a street if the two boundary chains from $s$ to $t$ are mutually weakly visible. We use a mobile robot to locate $t$ starting from $s$. Assume that the robot has a limited sensing capability that can only detect the constructed edges (also called gaps) on the boundary of its visible region, but cannot measure any angle or distance. The robot does not have knowledge of the street in advance. We present a new competitive strategy for this problem and prove that the length of the path generated by the robot is at most 9-times longer than the shortest path. We also propose a matching lower bound to show that our strategy is optimal. Compared with the previous strategy, we further relaxed the restriction that the robot should take a marking device and use the data structure S-GNT. The analysis of our strategy is tight.

Due to the nonlinear process of grain harvesting, there is no precise mathematical model to describe the behavior of the cleaning system of a harvester. Both the classical control and modern control methods cannot fulfil the requirements. Owing to this, the intelligent control algorithm was proposed, and the fuzzy logic control (FLC) method is a type of this. At present, most FLC algorithms are proposed in a MATLAB environment. However, the control problems in reality are controlled by microcomputer controllers with different chips. The control language of the microcomputer controller is usually written in C language. It is impossible to directly migrate the algorithm between these two different languages. Therefore, it is an important issue to transplant the FLC algorithm procedure written by MATLAB to the microcomputer controller. To realize the above target, we have built a complete set of control systems for our harvester’s cleaning system based on an upper computer and an STM32 core-chip controller. By means of combining FLC theory and expert knowledge, we adopted an improved FLC algorithm for the cleaning system, which is mounted in our self-designed combine harvester. Through this scheme, we have realized the objective of migrating the FLC algorithm from a MATLAB environment to the controller. The results of the experiment show that our method is reliable.