A Comprehensive Survey of Recent Transformers in Image,Video and Diffusion Models

Transformer models have emerged as dominant networks for various tasks in computer vision compared to Convolutional Neural Networks(CNNs).The transformers demonstrate the ability to model long-range dependencies by utilizing a self-attention mechanism.This study aims to provide a comprehensive survey of recent transformerbased approaches in image and video applications,as well as diffusion models.We begin by discussing existing surveys of vision transformers and comparing them to this work.Then,we review the main components of a vanilla transformer network,including the self-attention mechanism,feed-forward network,position encoding,etc.In the main part of this survey,we review recent transformer-based models in three categories:Transformer for downstream tasks,Vision Transformer for Generation,and Vision Transformer for Segmentation.We also provide a comprehensive overview of recent transformer models for video tasks and diffusion models.We compare the performance of various hierarchical transformer networks for multiple tasks on popular benchmark datasets.Finally,we explore some future research directions to further improve the field.

Ghost imaging based on the control of light source bandwidth

A scheme to improve the quality in ghost imaging(GI)by controlling the bandwidth of light source(BCGI)is proposed.The theoretical and numerical results show that the reconstruction result with high quality can be obtained by adjusting the bandwidth range of the light source appropriately,and the selection criterion of the bandwidth is analyzed by the power distribution of the imaging target.A proof-of-principle experiment is implemented to verify the theoretical and numerical results.In addition,the BCGI also presents better anti-noise performance when compared with some popular GI methods.

FPGA implementation and image encryption application of a new PRNG based on a memristive Hopfield neural network with a special activation gradient

A memristive Hopfield neural network(MHNN)with a special activation gradient is proposed by adding a suitable memristor to the Hopfield neural network(HNN)with a special activation gradient.The MHNN is simulated and dynamically analyzed,and implemented on FPGA.Then,a new pseudo-random number generator(PRNG)based on MHNN is proposed.The post-processing unit of the PRNG is composed of nonlinear post-processor and XOR calculator,which effectively ensures the randomness of PRNG.The experiments in this paper comply with the IEEE 754-1985 high precision32-bit floating point standard and are done on the Vivado design tool using a Xilinx XC7 Z020 CLG400-2 FPGA chip and the Verilog-HDL hardware programming language.The random sequence generated by the PRNG proposed in this paper has passed the NIST SP800-22 test suite and security analysis,proving its randomness and high performance.Finally,an image encryption system based on PRNG is proposed and implemented on FPGA,which proves the value of the image encryption system in the field of data encryption connected to the Internet of Things(Io T).

Evolution of defects and deformation mechanisms in different tensile directions of solidified lamellar Ti-Al alloy

Two-phaseγ-TiAl/α_(2)-Ti_(3)Al lamellar intermetallics have attracted considerable attention because of their excellent strength and plasticity.However,the exact deformation mechanisms remain to be investigated.In this paper,a solidified lamellar Ti-Al alloy with lamellar orientation at 0°,17°,and 73°with respect to the loading direction was stretched by utilizing molecular dynamics(MD)simulations.The results show that the mechanical properties of the sample are considerably influenced by solidified defects and tensile directions.The structure deformation and fracture were primarily attributed to an intrinsic stacking fault(ISF)accompanied by the nucleated Shockley dislocation,and the adjacent extrinsic stacking fault(ESF)and ISF formed by solidification tend to form large HCP structures during the tensile process loading at 73°.Moreover,cleavage cracking easily occurs on theγ/α_(2)interface under tensile deformation.The fracture loading mechanism at 17°is grain boundary slide whereas,at 73°and 0°,the dislocation piles up to form a dislocation junction.

Rucklidge-based memristive chaotic system:Dynamic analysis and image encryption

A new four-dimensional(4D)memristive chaotic system is obtained by introducing a memristor into the Rucklidge chaotic system,and a detailed dynamic analysis of the system is performed.The sensitivity of the system to parameters allows it obtains 16 different attractors by changing only one parameter.The various transient behaviors and excellent spectral entropy and C0 complexity values of the system can also reflect the high complexity of the system.A circuit is designed and verified the feasibility of the system from the physical level.Finally,the system is applied to image encryption,and the security of the encryption system is analyzed from multiple aspects,providing a reference for the application of such memristive chaotic systems.

Principle of subtraction ghost imaging in scattering medium

Scattering medium in light path will cause distortion of the light field,resulting in poor signal-to-noise ratio(SNR)of ghost imaging.The disturbance is usually eliminated by the method of pre-compensation.We deduce the intensity fluctuation correlation function of the ghost imaging with the disturbance of the scattering medium,which proves that the ghost image consists of two correlated results:the image of scattering medium and the target object.The effect of the scattering medium can be eliminated by subtracting the correlated result between the light field after the scattering medium and the reference light from ghost image,which verifies the theoretical results.Our research may provide a new idea of ghost imaging in harsh environment.

Heterogeneous Image Knowledge Driven Visual Perception

Dear Editor,This letter is concerned with visual perception closely related to heterogeneous images.Facing the huge challenge brought by different image modalities,we propose a visual perception framework based on heterogeneous image knowledge,i.e.,the domain knowledge associated with specific vision tasks,to better address the corresponding visual perception problems.

Object Helps U-Net Based Change Detectors

Dear Editor,This letter focuses on leveraging the object information in images to improve the performance of the U-Net based change detector.Change detection is fundamental to many computer vision tasks.Although existing solutions based on deep neural networks are able to achieve impressive results.

Sensitive Information Security Based on Elliptic Curves

The elliptic curve cryptography algorithm represents a major advancement in the field of computer security. This innovative algorithm uses elliptic curves to encrypt and secure data, providing an exceptional level of security while optimizing the efficiency of computer resources. This study focuses on how elliptic curves cryptography helps to protect sensitive data. Text is encrypted using the elliptic curve technique because it provides great security with a smaller key on devices with limited resources, such as mobile phones. The elliptic curves cryptography of this study is better than using a 256-bit RSA key. To achieve equivalent protection by using the elliptic curves cryptography, several Python libraries such as cryptography, pycryptodome, pyQt5, secp256k1, etc. were used. These technologies are used to develop a software based on elliptic curves. If built, the software helps to encrypt and decrypt data such as a text messages and it offers the authentication for the communication.

Artificial Intelligence Adoption for Cybersecurity in Africa

Legacy-based threat detection systems have not been able to keep up with the exponential growth in scope, frequency, and effect of cybersecurity threats. Artificial intelligence is being used as a result to help with the issue. This paper’s primary goal is to examine how African nations are utilizing artificial intelligence to defend their infrastructure against cyberattacks. Artificial intelligence (AI) systems will make decisions that impact Africa’s future. The lack of technical expertise, the labor pool, financial resources, data limitations, uncertainty, lack of structured data, absence of government policies, ethics, user attitudes, insufficient investment in research and development, and the requirement for more adaptable and dynamic regulatory systems all pose obstacles to the adoption of AI technologies in Africa. The paper discusses how African countries are adopting artificial intelligence solutions for cybersecurity. And it shows the impact of AI to identify shadow data, monitor for abnormalities in data access and alert cyber security professionals about potential threats by anyone accessing the data or sensitive information saving valuable time in detecting and remediating issues in real-time. The study finds that 69.16% of African companies are implementing information security strategies and of these, 45% said they use technologies based on AI algorithms. This study finds that a large number of African businesses use tools that can track and analyze user behaviour in designated areas and spot anomalies, such as new users, strange IP addresses and login activity, changes to permissions on files, folders, and other resources, and the copying or erasure of massive amounts of data. Thus, we discover that just 18.18% of the target has no national cybersecurity strategy or policy. The study proposes using big data security analytics to integrate AI. Adopting it would be beneficial for all African nations, as it provides a range of cyberattack defense techniques.

Review of Text Classification Methods on Deep Learning

Text classification has always been an increasingly crucial topic in natural language processing.Traditional text classification methods based on machine learning have many disadvantages such as dimension explosion,data sparsity,limited generalization ability and so on.Based on deep learning text classification,this paper presents an extensive study on the text classification models including Convolutional Neural Network-Based(CNN-Based),Recurrent Neural Network-Based(RNN-based),Attention Mechanisms-Based and so on.Many studies have proved that text classification methods based on deep learning outperform the traditional methods when processing large-scale and complex datasets.The main reasons are text classification methods based on deep learning can avoid cumbersome feature extraction process and have higher prediction accuracy for a large set of unstructured data.In this paper,we also summarize the shortcomings of traditional text classification methods and introduce the text classification process based on deep learning including text preprocessing,distributed representation of text,text classification model construction based on deep learning and performance evaluation.

Decentralized Attribute-Based Encryption and Data Sharing Scheme in Cloud Storage

In this paper, we consider the problems of data sharing between multiple distrusted authorities. Prior solutions rely on trusted third parties such as CAs, or are susceptible to collusion between malicious authorities, which can comprise the security of honest ones. In this paper, we propose a new multi-authority data sharing scheme – Decentralized Multi-Authority ABE(DMA), which is derived from CP-ABE that is resilient to these types of misbehavior. Our system distinguishes between a data owner(DO) principal and attribute authorities(AAs): the DO owns the data but allows AAs to arbitrate access by providing attribute labels to users. The data is protected by policy encryption over these attributes. Unlike prior systems, attributes generated by AAs are not user-specific, and neither is the system susceptible to collusion between users who try to escalate their access by sharing keys. We prove our scheme correct under the Decisional Bilinear Diffie-Hellman(DBDH) assumption; we also include a complete end-to-end implementation that demonstrates the practical efficacy of our technique.

Robust Watermarking of Mobile Video Resistant against Barrel Distortion

In head mounted display(HMD),in order to cancel pincushion distortion,the images displayed on the mobile should be prewarped with barrel distortion.The copyright of the mobile video should be verified on both the original view and the pre-warped virtual view.A robust watermarking resistant against barrel distortion for HMDs is proposed in this paper.Watermark mask is embedded into image in consideration of imperceptibility and robustness of watermarking.In order to detect watermark from the pre-warped image with barrel distortion,an estimation method of the barrel distortion is proposed for HMDs.Then,the same warp is enforced on the embedded watermark mask with the estimated parameters of barrel distortion.The correlation between the warped watermark and the pre-warped image is computed to predicate the existence of watermark.As shown in experimental results,watermark of mobile video can be detected not only from the original views,but also from the pre-warped virtual view.It also shows that the proposed scheme is resistant against combined barrel distortion and common post-processing,such as JPEG compression.

Reliability analysis of monotone coherent multi-state systems based on Bayesian networks

The Bayesian networks (BNs) provide a robust probabilistic method of reasoning under uncertainty and have been successfully applied to a variety of real-world tasks. Aiming to explore the capabilities of the BN formalism in reliability analysis of monotone coherent multi-state systems, the BNs are compared with a popular tool for reliability analysis of monotone coherent multi-state systems, namely the multi-state fault trees (MFTs). It is shown that any MFT can be directly mapped into BN and the basic inference techniques on the latter may be used to obtain classical parameters computed from the former (i.e. probability distribution of top variable, minimal upper vectors and maximum lower vectors for any performance level, importance measures of components). Furthermore, some additional information can be obtained by using BN, both at the modeling and analysis level. At the modeling level, several restrictive assumptions implicit in the MFT methodology can be removed and various kinds of dependencies among components can be accommodated. At the analysis level, a general diagnostic analysis can be performed. The comparison of these methods is illustrated by an example of the water supply system. © 2016 Beijing Institute of Aerospace Information.

Research on Data Fusion of Adaptive Weighted Multi-Source Sensor

Data fusion can effectively process multi-sensor information to obtain more accurate and reliable results than a single sensor.The data of water quality in the environment comes from different sensors,thus the data must be fused.In our research,self-adaptive weighted data fusion method is used to respectively integrate the data from the PH value,temperature,oxygen dissolved and NH3 concentration of water quality environment.Based on the fusion,the Grubbs method is used to detect the abnormal data so as to provide data support for estimation,prediction and early warning of the water quality.

A blockchain-based trustworthy collaborative power trading scheme for 5G-enabled social internet of vehicles

Social Internet of Vehicles(SIoV)falls under the umbrella of social Internet of Things(IoT),where vehicles are socially connected to other vehicles and roadside units that can reliably share information and services with other social entities by leveraging the capabilities of 5G technology,which brings new opportunities and challenges,e.g.,collaborative power trading can address the mileage anxiety of electric vehicles.However,it relies on a trusted central party for scheduling,which introduces performance bottlenecks and cannot be set up in a distributed network,in addition,the lack of transparency in state-of-the-art Vehicle-to-Vehicle(V2V)power trading schemes can introduce further trust issues.In this paper,we propose a blockchain-based trustworthy collaborative power trading scheme for 5G-enabled social vehicular networks that uses a distributed market mechanism to introduce trusted power trading and avoids the dependence on a centralized dispatch center.Based on the game theory,we design the pricing and trading matching mechanism for V2V power trading to obtain maximum social welfare.We use blockchain to record power trading data for trusted pricing and use smart contracts for transaction matching.The simulation results verify the effectiveness of the proposed scheme in improving social welfare and reducing the load on the grid.

Multi-Label Learning Based on Transfer Learning and Label Correlation

In recent years,multi-label learning has received a lot of attention.However,most of the existing methods only consider global label correlation or local label correlation.In fact,on the one hand,both global and local label correlations can appear in real-world situation at same time.On the other hand,we should not be limited to pairwise labels while ignoring the high-order label correlation.In this paper,we propose a novel and effective method called GLLCBN for multi-label learning.Firstly,we obtain the global label correlation by exploiting label semantic similarity.Then,we analyze the pairwise labels in the label space of the data set to acquire the local correlation.Next,we build the original version of the label dependency model by global and local label correlations.After that,we use graph theory,probability theory and Bayesian networks to eliminate redundant dependency structure in the initial version model,so as to get the optimal label dependent model.Finally,we obtain the feature extraction model by adjusting the Inception V3 model of convolution neural network and combine it with the GLLCBN model to achieve the multi-label learning.The experimental results show that our proposed model has better performance than other multi-label learning methods in performance evaluating.

The dynamics of a memristor-based Rulkov neuron with fractional-order difference

The exploration of the memristor model in the discrete domain is a fascinating hotspot.The electromagnetic induction on neurons has also begun to be simulated by some discrete memristors.However,most of the current investigations are based on the integer-order discrete memristor,and there are relatively few studies on the form of fractional order.In this paper,a new fractional-order discrete memristor model with prominent nonlinearity is constructed based on the Caputo fractional-order difference operator.Furthermore,the dynamical behaviors of the Rulkov neuron under electromagnetic radiation are simulated by introducing the proposed discrete memristor.The integer-order and fractional-order peculiarities of the system are analyzed through the bifurcation graph,the Lyapunov exponential spectrum,and the iterative graph.The results demonstrate that the fractional-order system has more abundant dynamics than the integer one,such as hyper-chaos,multi-stable and transient chaos.In addition,the complexity of the system in the fractional form is evaluated by the means of the spectral entropy complexity algorithm and consequences show that it is affected by the order of the fractional system.The feature of fractional difference lays the foundation for further research and application of the discrete memristor and the neuron map in the future.

Toward Energy-Efficiency Optimization of Pktgen-DPDK for Green Network Testbeds

The packet generator(pktgen) is a fundamental module of the majority of soft ware testers used to benchmark network pro tocols and functions. The high performance of the pktgen is an important feature of Future Internet Testbeds, and DPDK is a network packet accelerated platform, so we can use DPDK to improve performance. Meanwhile green computing is advocated for in the fu ture of the internet. Most existing efforts have contributed to improving either performance or accuracy. We, however, shifted the focu to energy-efficiency. We find that high per formance comes at the cost of high energy consumption. Therefore, we started from a widely used high performance schema, deeply studying the multi-core platform, especially in terms of parallelism, core allocation, and fre quency controlling. On this basis, we proposed an AFfinity-oriented Fine-grained CONtrolling(AFFCON) mechanism in order to improve energy efficiency with desirable performance As clearly demonstrated through a series o evaluative experiments, our proposal can re duce CPU power consumption by up to 11%while maintaining throughput at the line rate.

A Secure Mutual Authentication Scheme of Blockchain-Based in WBANs

Wireless Body Area Networks(WBANs)refer to small sensor network that consists of sensor devices mounted on the surface of the body or implanted in the body,as such networks are employed to harvest physiological data of the human body or to act as an assistant regulator of several specific physiological indicators of the human body.The sensor devices transmit the harvested human physiological data to the local node via a public channel.Before transmitting data,the sensor device and the local node should perform mutual authentication and key agreement.It is proposed in this paper a secure mutual authentication scheme of blockchain-based in WBANs.To analyze the security of this scheme,formal security analysis,and informal security analysis are used,then the computation and communication costs are compared with those of the relevant schemes.Relevant experimental results reveal that the proposed scheme exhibit more effective control over energy consumption and promising.