Network embedding aims to represent nodes with low dimensional vectors while preserving structural information. It has been recently shown that many popular network embedding methods can be transformed into matrix factorization problems. In this paper, we propose the unifying framework “Z-NetMF,” which generalizes random walk samplers to Z-Laplacian graph filters, leading to embedding algorithms with interpretable parameters. In particular, by controlling biases in the time domain, we propose the Z-NetMF-t algorithm, making it possible to scale contributions of random walks of different length. Inspired by node2vec, we design the Z-NetMF-g algorithm, capturing the random walk biases in the graph domain. Moreover, we evaluate the effect of the bias parameters based on node classification and link prediction tasks. The results show that our algorithms, especially the combined model Z-NetMF-gt with biases in both domains, outperform the state-of-art methods while providing interpretable insights at the same time. Finally, we discuss future directions of the Z-NetMF framework.

Near-infrared (NIR) images and visible (VIS) images can provide mutually complementary information for each other, thus the fusion of the two modalities can create images of high quality even in adverse conditions. However, the luminance of NIR and VIS images may be inconsistent in some regions, resulting in color distortion and unrealistic appearance in the fused images. The existing methods perform poorly at luminance retention. Aiming at the problem and based on deep learning framework, we propose an edge-guided method which can be applied to the image fusion network. Edge maps are utilized as prior knowledge of images to boost the performance of the neural network. Additionally, we propose a luminance-preserving loss function combined with max-edge loss to further improve the image quality. Experimental results show the superiority of our method.

Federated learning is designed to train models in a distributed scheme while keeping the clients' data stored locally. The aggregation server only receives local models from clients and does not require clients to upload their local data, in which way it protects the clients' privacy. However, federated learning is vulnerable. The federated learning models are sensitive to poisoning attacks. Existing data poisoning attack methods assume that the attacker and the client have the same data distribution and data volume, which is unpractical. In this paper, we first propose a privacy inference-based poisoning data generation method, FLPDG. FLPDG changes the relationship between data and labels, and uses the data of benign clients to launch poisoning attacks. This method relies on the global model of an iterative update to obtain the data and labels of benign clients. Second, a privacy inference-based data poisoning attack model Poi_PDG is proposed. This model uses the FLPDG method to launch a data poisoning attack under conditions of insufficient original data volume of the attacker. Meanwhile, a defense method PDG_DF is proposed for Poi_PDG. It splits the image data into variance regions and utilizes GANs to hide the visual features of each image region. It controls the degree of feature hiding by setting different thresholds to keep the classification features of the image while ensuring the accuracy of the training model. Finally, several experiments are conducted to evaluate the proposed attack and defense methods, and the experimental results indicate the effectiveness of the methods.

Since deep learning was introduced into style transfer, remarkable results have been achieved in it and it is widely used in multimedia fields, such as photography. However, the computational costs of the existing state-of-the-art (SOTA) arbitrary style transfer algorithms are still too complex to apply them on mobile device and high resolution, and their performance on shape retention is not satisfactory enough. To deal with the above problems, we propose a novel arbitrary style transfer algorithm. Specially, we propose a new network which ensures the low computational cost and high shape retention. Moreover, we propose the weighted style loss function to improve the performance on style migration. The experimental results show that the proposed algorithm achieves better results with lower computational cost than the SOTA algorithms.

This paper presents a solution to the real-time detection of MAC layer misbehaviors in IEEE 802.11 networks. Among the wide range of misbehaviors, we focus on the sender side selfish behavior that creates a channel- capturing effect by using favorable parameters, and the receiver side selfish behavior that does not respond with CTS and ACK upon receiving RTS and data packets, which clears the channel for itself and causes its sender to waste resources. These misbehaviors are subtle to detect, and yet can undermine the performance of the well-behaved nodes significantly. This paper shows a powerful real-time detection method that can catch these misbehaviors as soon as they have started. The detection method requires collecting delay, throughput, and packet interval data to generate time series and applying a sequential change point detection algorithm on the data streams as soon as new data points come in. All attacks are simulated in ns-3 and the simulation results verified the effectiveness of the detection method.

Due to the open nature of wireless communication medium, wireless networks are susceptible to jamming attacks. Jammers interfere with the legitimate nodes by sending strong jamming signals. Legitimate nodes can successfully transmit only between the gaps of the jamming signals. It is therefore very important to detect a jamming attack as soon as it happens in order to effectively take counter measurements. There are various types of jamming attacks, however, the {\it signature} of all jamming attacks is the performance degradation of legitimate nodes. Based on this observation, we develop a detection method using time series analysis approach. We model the network measurements taken over time as time series, and employ a sequential change point detection algorithm to detect the change of state in the time series, which is an indicator of change in the network state. Timely and accurate detection is the first step before further identification and localization of the source of interference. In this paper, we address the detection part and leave the localization of the jammer to future work. The jamming attacks are simulated in ns-3 simulator, and the detection result is satisfactory in terms of false alarm rate and detection delay.

This paper addresses detecting in-band wormholes in wireless ad hoc networks. The detection scheme requires collecting the end-to-end delay of packets at the receiver and then applying a sequential change point detection algorithm to detect abrupt changes in the delay time series. A new change point detection algorithm, named SW-CLT, is proposed. The algorithm is based on the Central Limit Theorem (CLT) and does not involve using a preset detecting threshold. The algorithm is compared with the non-parametric cumulative sum (NP-CUSUM) because the non-parametric version is believed to be more robust to highly dynamic data than the parametric version. SW-CLT has the ability to adjust its detection threshold with the variance of the data, and therefore is more robust than NP-CUSUM, which uses a preset threshold. Simulation results from ns3 verified the advantage of SW-CLT over NP-CUSUM in all simulated scenarios.

We address the issue of wireless ad hoc network connectivity by using a tail model that is derived from the signal to interference and noise ratio (SINR). The SINR model more accurately describes link connectivity than the traditionally used disk model in the real-world. We first assess the network connectivity by measuring the conductance of the network and find the bottleneck location of the network, and then deploy a relay node to improve the connectivity at the bottleneck. A partition algorithm is proposed to address the first problem, and an optimization problem is proposed to address the relay node deployment problem. The relay node deployment problem is solved by using approximate convex optimization models, and the approximation performance is analyzed. Simulation results show that the partition algorithm based on the SINR model identifies the network bottleneck more accurately than the previous methods based on the binary model. It also verifies that the relay node can significantly relieve the bottleneck and make the network more tightly knit.

We consider the network connectivity problem in a wireless ad hoc network. Network connectivity is measured by the conductance of the network, also called the Cheeger constant of the graph. A partition algorithm based on this measure is developed that divides the network at the bottleneck area. After the network is bisected, a relay node may be deployed between the two parts to increase the conductance of the network. The relay node deployment problem is formulated as an integer linear program to maximize the number of connections between the nodes on the two sides of the cut, and then a convex optimization algorithm is used to find the precise location of the relay node, which is within the convex hull defined by the radio transmission ranges of all the nodes that can connect to the relay node. The relay node significantly relieves the bottleneck, and the graph connectivity measured by other metrics such as the widely used Fiedler value are also increased.

This research collected the data of students' vision tests in Zhoushan and other regions in China, made a quantitative analysis of the students' vision and describes the relationship between vision and seafarers' Medical Examination System, as well as other safety issues concerned with vision. The statistics showed that there are about 65% students with poor vision in high schools and almost 85% students with poor vision in colleges, which will be a risk to maritime human resources in future. Finally, some comments and recommends are given for improving students' vision.