With the rapid development of generative artificial intelligence and image editing technology, more and more fake images, such as face forgery and partial tampering, are appearing, posing a serious threat to information security and the social trust system. To address the problems of strong dependence on single spatial-domain features, poor generalisation across datasets, and unexplainable detection results, this paper proposes a forged-image detection system based on dual-stream fusion networks in the spatial and frequency domains. The accuracy, robustness, and interpretability of forged image detection are improved across both the algorithm and the system. At the algorithmic level, this paper proposes a spatial-frequency-domain dual-stream feature-extraction and fusion architecture. Firstly, a frequency-domain analysis module based on the fast Fourier transform (FFT) is established, which can detect resampling traces and spectral anomalies that cannot be detected in the spatial domain. At the system level, the FastAPI framework is used to engineer the deployment of the detection system. The system integrates modules for image preprocessing, two-stream network reasoning, and result visualisation. It can accept the user’s image upload request, perform real-time image detection, and clearly show the authenticity probability and a tampering heat map. It has good interactivity and application value. In this study, mainstream datasets such as FaceForensics++, Stable Diffusion-generated images, and GreatSplicing were selected for comprehensive verification. The experimental results show that the spatial-frequency domain dual-stream fusion architecture can achieve detection accuracy greater than 90

There are various indoor fingerprint localization techniques utilizing the similarity of received signal strength (RSS) to discriminate the similarity of positions. However, due to the varied states of different wireless access points (APs), each AP’s contribution to RSS similarity varies, which affects the accuracy of localization. In our study, we analyzed several critical causes that affect APs’ contribution, including APs’ health states and APs’ positions. Inspired by these insights, for a large-scale indoor space with ubiquitous APs, a threshold was set for all sample RSS to eliminate the abnormal APs dynamically, a correction quantity for each RSS was provided by the distance between the AP and the sample position to emphasize closer APs, and a priority weight was designed by RSS differences (RSSD) to further optimize the capability of fingerprint distances (FDs, the Euclidean distance of RSS) to discriminate physical distance (PDs, the Euclidean distance of positions). Integrating the above policies for the classical WKNN algorithm, a new indoor fingerprint localization technique is redefined, referred to as FDs’ discrimination capability improvement WKNN (FDDC-WKNN). Our simulation results showed that the correlation and consistency between FDs and PDs are well improved, with the strong correlation increasing from 0 to 76% and the high consistency increasing from 26% to 99%, which confirms that the proposed policies can greatly enhance the discrimination capabilities of RSS similarity. We also found that abnormal APs can cause significant impact on FDs discrimination capability. Further, by implementing the FDDC-WKNN algorithm in experiments, we obtained the optimal K value in both the simulation scene and real library scene, under which the mean errors have been reduced from 2.2732 m to 1.2290 m and from 4.0489 m to 2.4320 m, respectively. In addition, compared to not using the FDDC-WKNN, the cumulative distribution function (CDF) of the localization errors curve converged faster and the error fluctuation was smaller, which demonstrates the FDDC-WKNN having stronger robustness and more stable localization performance.

The weighted K-nearest neighbors (WKNN) algorithm has been widely used in indoor fingerprint localization, which utilizes fingerprint distance (FD) of pairwise positions to discriminate the physical distance (PD) between them. However, due to the varied states of different wireless access points (APs), their contribution to FD discrimination capability varies. In our study, we analyzed several critical causes that affect APs’ contribution, including AP’s health state, the distance, and the direction between APs and position pairs. Inspired by these insights, a threshold was set for all sample RSS to eliminate the impact of abnormal APs on FD, and a correction quantity for each RSS was provided by the distance between APs and sample positions to reduce the strong signal influence on FD. Furthermore, a priority weight was designed by RSS differences (RSSD) to further optimize FD’s capability to discriminate PD. Integrating the above policies, a new indoor fingerprint localization technique is redefined, referred to FD’s discrimination capability improvement WKNN (FDDC-WKNN), which is suitable for indoor scenes with a large number of APs that are not uniformly managed. Our simulation results show that the correlation and consistency between FD and PD are well improved, with strong correlation increasing from 0 to 76% and high consistency increasing from 26% to 99%, which confirms that the proposed policies can greatly enhance FD’s discrimination capabilities to PD, and we also found that abnormal APs can cause significant impact on FDs’ discrimination Capability. Further, by implementing the FDDC-WKNN algorithm in experiments, we obtained the optimal K value in both the simulation scene and real library scene, under which the average localization errors have been reduced from 2.2732m to 1.2183m and from 3.4295m to 2.2068m, respectively. In addition, compared to not using the FDDC-WKNN, the cumulative distribution function (CDF) of the localization errors curve converged faster and the errors fluctuation is smaller, which demonstrate FDDC-WKNN having stronger robustness and more stable localization performance to the state of APs in indoor environments.

Vehicle re-identification is a technology that continuously tracks and identifies vehicles in different spatial domains, and playing a critical role in Space-Air-Ground-Sea Integrated Networks(SAGSIN). Viewpoint variation problem, that is vehicle appearance changes greatly under various viewpoints, makes vehicle re-identification challenging. To eliminate the negative effects of viewpoint variation, in this paper, we propose a Multi-View Vehicle Image Generation Network for Vehicle Re-Identification(MVIGN). MVIGN generate images with the same identity as the input vehicle image but with a different and controllable pose to solve viewpoint variation problem. Extensive experiments indicate using images generated by MVIGN to expand training set can improve the model accuracy and reduce the cost of manually collecting and labeling data.

The Internet of Vehicles (IoV) enables vehicles to share data that help vehicles perceive the surrounding environment. However, vehicles can spread false information to other IoV nodes; this incorrect information misleads vehicles and causes confusion in traffic, therefore, a vehicular trust model is needed to check the trustworthiness of the message. To eliminate the spread of false information and detect malicious nodes, we propose a double-layer blockchain trust management (DLBTM) mechanism to objectively and accurately evaluate the trustworthiness of vehicle messages. The double-layer blockchain consists of the vehicle blockchain and the RSU blockchain. We also quantify the evaluation behavior of vehicles to show the trust value of the vehicle's historical behavior. Our DLBTM uses logistic regression to accurately compute the trust value of vehicles, and then predict the probability of vehicles providing satisfactory service to other nodes in the next stage. The simulation results show that our DLBTM can effectively identify malicious nodes, and over time, the system can recognize at least 90% of malicious nodes.

In some security domains, the defender holds more private information than the attacker, such as the authenticity of resources. A mock camera can be adopted by the defender to deter potential attackers while the attacker may see it as a real one. This provides an opportunity for a defender with fake resources to bluff, as if all resources she owns are real. We model the bluffing behavior as a signaling game where the attacker is uncertain whether the defender is strong (all resources are real) or weak (fake resources disguised as real resources). We propose a mixed integer cubic programming formulation to solve for the solution concept based on perfect Bayesian equilibrium. We compare our bluffing security game model with a model without signaling. The experiments indicate that our model outperforms the no-signaling model in terms of the defender's utility.

The recommendation services are critical for IoT since they provide interconnection between various devices and services. In order to make Internet searching convenient and useful, algorithms must be developed that overcome the shortcomings of existing online recommendation systems. Therefore, a novel Stein Variational Recommendation System algorithm (SVRS) is proposed, developed, implemented and tested in this paper in order to address the long-standing recommendation problem. With Stein’s identity, SVRS is able to calculate the feature vectors of users and ratings it has generated, as well as infer the preference for users who have not rated certain items. It has the advantages of low complexity, scalability, as well as providing insights into the formation of ratings. A set of experimental results revealed that SVRS performed better than other types of recommendation methods in root mean square error (RMSE) and mean absolute error (MAE).

This paper proposes a verification method of power-off protection mechanism for non-volatile memory, which is suitable for the application scenarios where the data needs to be protected when power-off occurs when the non-volatile memory is erased or written. This method can verify the correctness of the power-off protection mechanism, i.e. whether the data operated after the power-off event can maintain atomicity. This verification method gives the specific verification method, but does not limit the specific details of the implementation.

With the advancement of science and the emergence of new technologies, technology opportunities analysis has attracted increasing attention from both society and academia. This study proposes a hybrid approach to integrate topic modeling, semantic SAO analysis, machine learning, and expert judgment, identifying technological topics and potential development opportunities. The systematical methodology is applied to analyze a set of 9,883 Derwent Innovation Index (DII) patents related to the dye-sensitized solar cell to present its potential contribution of technical intelligence for R&D management. Also, how the approach is validated and optimized is illustrated. The main contributions of this paper are two-fold. First, an optimized topic extraction model with high accuracy is constructed, considering both the patent classification codes and term location. Second, we integrate the topic modeling, SAO technique, and machine learning to explore semantic relationships among technological topics represented as a suite of terms. The methodology overcomes some drawbacks of the current studies. It can be used as a powerful tool for technological opportunities analysis.

Activity selection is critical for the smart environment and Cyber-Physical Systems (CPSs) that can provide timely and intelligent services, especially as the number of connected devices is increasing at an unprecedented speed. As it is important to collect labels by various agents in the CPSs, crowdsourcing inference algorithms are designed to help acquire accurate labels that involve high-level knowledge. However, there are some limitations in the algorithm in the existing literature such as incurring extra budget for the existing algorithms, inability to scale appropriately, requiring the knowledge of prior distribution, difficulties to implement these algorithms, or generating local optima. In this paper, we provide a crowdsourcing inference method with variational tempering that obtains ground truth as well as considers both the reliability of workers and the difficulty level of the tasks and ensure a local optimum. The numerical experiments of the real-world data indicate that our novel variational tempering inference algorithm performs better than the existing advancing algorithms. Therefore, this paper provides a new efficient algorithm in CPSs and machine learning, and thus, it makes a new contribution to the literature.

With the rapid development of modern network technology, many new technologies have sprung up in the era of big data, which means that Internet finance has been an indelible trend from its appearance to its gradual maturity. Of course, Internet finance has developed to the present, behind not only the efforts of the people, but also the continuous support and encouragement of the government and the state in policy. As a result, the “financial + Internet” model in the public unremitting exploration and innovation, showing an increasingly vigorous face. However, this has caused some conflict and competition to the development of traditional commercial banks. Based on the concept, characteristics and models of traditional commercial banks and Internet finance, this paper studies the competition and development of Internet finance to traditional commercial banks from the aspects of online payment, customer lending, personal finance and so on. It can be seen from this study that traditional commercial banks can make a better transformation with the help of the Internet, and at the same time, Internet finance can better solve their own risks and further improve the service system with the help of traditional commercial banks. This shows that there is a mutually reinforcing and mutually restricted relationship between the two.

The human action recognition technology has developed rapidly in recent years. The technologies of RNN and 3D convolution based on posture information and video frame information respectively have achieved high accuracy using various data sets, however, both of them have shortcomings in the field of abnormal behavior recognition. The definition of abnormal behavior needs to consider not only the action type simply, but also the environmental information comprehensively, so there are limitations in using RNN only based on posture information. Due to the input characteristics, action recognition technology based on 3D convolution is more related to environmental information and group behavior information, it cannot locate the action time accurately. This paper proposed an abnormal behavior recognition framework based on P3D and LSTM. The framework used pre-trained P3D to extract environmental features, and adopted pre-trained LSTM to extract individual action features to help system for time positioning, finally apply ranking model to classify abnormal behaviors after combining environmental features with action features. When training LSTM model, a regression network was added to enhance its time positioning ability. The experiment showed that the proposed framework based on P3D and LSTM has a greater improvement in the recognition accuracy and time positioning than only using 3D convolution technology or LSTM technology, and can accurately recognize abnormal behaviors.

The Internet of Things (IoT) interconnects various devices and services, of which recommendation services are an important component to help the development of IoT applications. Furthermore, without the aid of suitable online recommendation systems, Internet users will be overwhelmed by the tremendous amount of contents. Researchers have thus developed a large volume of recommendations. However, they are all flawed with high complexity, cold start issues, inability to generalize, etc. In recent years, some researchers had turned to variational inference (VI)-based recommendation systems, which can solve the above problems to some extent. However, these VI-based recommendations are merely hybrid methods of VI with the existing recommendation algorithms and are unable to be implemented well in real practices. Therefore, developing algorithms that can overcome these limitations of the existing online recommendation systems is essential for convenient and useful Internet searches. In this paper, we propose, develop, implement and test a more general, new and innovative Stein Variational Recommendation System algorithm (SVRS) to tackle the long plaguing recommendation problems. Based on Stein’s identity, the SVRS algorithm can compute the feature vector of existing users and items it had rated, and further predict the ratings for users that have not been engaged with certain content. SVRS provides more general insights into the forming of user ratings, can be easily extended to higher dimensions and has the merits of low complexity, easy scaling and generalizability. Experiments show that SVRS outperforms the other existing type of recommendation algorithms and it has higher accuracy in terms of mean absolute error (MAE) and root mean square error (RMSE).

In this article, we explore the use of multivariate Archimedean copulas in modelling the mortality dependence between different countries and pricing mortality bonds. We study the fitting performance of multi-dimensional, fully nested, and partially nested Archimedean copulas and test 11 types of generators and two skewed distributions. To evaluate their practical usefulness, we adopt the fitted models to compute the market prices for some typical mortality bond structures. The results show that the copula assumption has a significant impact on the calculation of the prices of mortality-linked securities and the management of extreme mortality risks.

With the rapid development of wireless sensor networks (WSNs) technology, a growing number of applications and services need to acquire the states of channels or sensors, especially in order to use these states for monitoring, object tracking, motion detection, etc. A critical issue in WSNs is the ability to estimate the source parameters from the readings of a distributed sensor network. Although there are several studies on channel estimation (CE) algorithms, existing algorithms are all flawed with their high complexity, inability to scale, inability to ensure the convergence to a local optimum, low speed of convergence, etc. In this work, we turn to variational inference (VI) with tempering to solve the channel estimation problem due to its ability to reduce complexity, ability to generalize and scale, and guarantee of local optimum. To the best of our knowledge we are the first to use VI with tempering for advanced channel estimation. The parameters that we consider in the channel estimation problem include pilot signal and channel coefficients, assuming there is orthogonal access between different sensors (or users) and the data fusion center (or receiving center). By formulating the channel estimation problem into a probabilistic graphical model, the proposed Channel Estimation Variational Tempering Inference (CEVTI) approach can estimate the channel coefficient and the transmitted signal in a low-complexity manner while guaranteeing convergence. CEVTI can find out the optimal hyper-parameters of channels with fast convergence rate, and can be applied to the case of code division multiple access (CDMA) and uplink massive multi-input-multi-output (MIMO) easily. Simulations show that CEVTI has higher accuracy than state-of-the-art algorithms under different noise variance and signal-to-noise ratio. Furthermore, the results show that the more parameters are considered in each iteration, the faster the convergence rate and the lower the non-degenerate bit error rate with CEVTI. Analysis shows that CEVTI has satisfying computational complexity, and guarantees a better local optimum. Therefore, the main contribution of the paper is the development of a new efficient, simple and reliable algorithm for channel estimation in WSNs.

The traditional collaborative filtering recommendation algorithm has the problem of data sparsity and expansibility. Aiming at this problem, and improved bisecting k-means collaborative filtering algorithm proposed.The algorithm first removes unrated items in the rating data matrix based on the Weighted Slope One algorithm preprocessing to reduce its sparsity. Then the preprocessed rating data is clustered based on the bisecting K-means algorithm, which reduces the nearest neighbor search space of the target user by assembling similar objects, thereby improving the algorithm’s expansibility. Finally, use the recommendation algorithm to generate the final result.Experimental results show that the improved bisecting k-means algorithm improves the recommendation effect.

Background: The National Natural Science Foundation of China (NSFC) is a vital source of support for scientific research in China and abroad. Among the 18 categories programs managed by the NSFC, the Joint Fund, which looks to develop innovation in basic Chinese research, has drawn particular interest from scientists and researchers. This study set out to review the distribution of funding from the Joint Fund in recent years to provide potential applicants with a useful point of reference. Methods: Guidebooks and annual reports published from 2015 to 2019 were accessed on the NSFC website and the relevant data from the Joint Fund were obtained. The data were analyzed to identify any trends or patterns in funding distribution. Results: Between 2015 and 2019, the NSFC's Joint Fund awarded a total of 46 million Chinese yuan (CNY) to 354 projects in 29 areas within the field of population health. The funding recipients, who were aged between 30 and 66 years old, came from 47 host institutions. Among the successful applicants, 225 were in senior level positions, 129 were in deputy senior level positions, and 19 were in medium-grade professional positions. Collaboration between institutions was a common feature of funding applications. Conclusions: By analyzing the distribution of funding from the Joint Fund, we identified a number of common features of successful funding applications, which may help to inform future funding applications.

We study the so-called best-performance life expectancy from a theoretical perspective by using a generalised theorem of the extreme value distribution. Under the generalised extreme value theorem, we experiment with two mathematical models called copulas to model the relationships between countries, and also test alternative parameterisations to co-model the data of both sexes, incorporating both common and sex-specific parameters. In earlier work on using the extreme value distribution for best-performance life expectancy, all countries were assumed to be independent, and the two sexes were treated separately. In this paper, we apply the generalised extreme value theorem with copulas in order to relax the independence assumption between life expectancies of different countries under extreme events. We find that the two resulting limiting distributions model the data reasonably well and allow us to make future projections with probabilistic intervals. Setting certain common parameters between females and males can also improve the information content and parameter parsimony. In particular, we detect significant dependence between countries and notice that the dependence tends to be stronger for males. Based on the aggregate experience of numerous countries, the projected trends may serve as a reference or an upper bound for life expectancy forecasts of individual countries.

The strong, persistent linear trend of the highest period life expectancy of females at birth, namely best-performance life expectancy, is an interesting global phenomenon which has already lasted for about 150years. In this article, we study both the highest and lowest period life expectancies of a group of more developed countries and areas, and exploit their recent trends to construct approximate upper and lower bounds as a supplementary tool for future projections. We also seek two modifications of this proposed approach. First, despite that it has remained largely an empirical observation, we intend to examine the use of extreme value theory to provide a more theoretical framework for both the highest and lowest life expectancies. Second, we construct two hypothetical populations with each age experiencing the lowest or highest mortality rate amongst all the populations considered, and extrapolate their life expectancy trends into the future. The resulting life expectancy bounds perform reasonably well in our backtesting exercise and can potentially complement the usual application of a stochastic mortality model to the data of a single country.

The power equipment assets has the characteristics of large initial investment, high operation cost, long service time. Therefore it is the basic requirement to carry out the electrical equipment life cycle management to ensure the grid assets safety running. Using security Ultra High Frequency (UHF) Radio Frequency Identification (RFID) anti metal tag as the equipment asset identification, it is useful to ensure the legitimacy of reader and tag, and the security of data storage and transmission, and the personnel authority access control. In this paper, the characteristics, working principle and working flow of high security RFID electronic tags are designed and studied. The typical solution is provided for the intelligent management of grid assets.