The proliferation of open-source Large Language Models (LLMs) underscores the pressing need for evaluation methods. Existing works primarily rely on external evaluators, focusing on training and prompting strategies. ...
详细信息
Despite ongoing efforts to defend neural classifiers from adversarial attacks, they remain vulnerable, especially to unseen attacks. In contrast, humans are difficult to be cheated by subtle manipulations, since we ma...
作者：
Chen, LiyanFeng, BingwenXia, ZhihuaLu, WeiWeng, JianJinan University
College of Cyber Security Guangzhou510632 China Jinan University
College of Cyber Security Engineering Research Center of Trustworthy AI Ministry of Education Guangzhou510632 China Sun Yat-sen University
School of Computer Science and Engineering Guangdong Province Key Laboratory of Information Security Technology Ministry of Education Key Laboratory of Machine Intelligence and Advanced Computing Guangzhou510006 China
Generative steganography stands as a promising technique for information hiding, primarily due to its remarkable resistance to steganalysis detection. Despite its potential, hiding a secret image using existing genera...
详细信息
While playing an increasing role in the field of air-space integrated networks,terminal entities are exposed to more serious security risks than ordinary terminal entities on the ground,including but not limited to as...
详细信息
While playing an increasing role in the field of air-space integrated networks,terminal entities are exposed to more serious security risks than ordinary terminal entities on the ground,including but not limited to astronomical risks(e.g.,solar activity),link disruptions,and malicious *** the integrated air-space-space network,after the terminal entity is loaded into the rocket for launch,real-time monitoring measures are implemented on the ground site for all links to the intended *** top of the real-time monitoring measures,the need for a comprehensive evaluation of the air-space integration network is *** this paper,we further classify the security requirements and evaluation benchmarks of the terminal entities of the air-space integration network around the security evaluation requirements of the air-space integration *** this basis,the relevant concept of trust value is introduced to assess the security status of terminal entities of air-space-sky integrated networks,and on this basis,a security assessment system of air-space-sky integrated network terminals is *** exploration and development of field applications are initially realized.
In recent years, with the continuous progress of deep learning neural networks and the construction of large data sets, new breakthroughs in voiceprint verification technology for voice calls have been rapidly achieve...
详细信息
A new cyclic prefix(CP)-based nonoverlapping FBMC-QAM(CP-NO-FBMC-QAM)system with two prototype filters is proposed in this paper,which satisfies complex orthogonality conditions and good frequency energy confinement a...
详细信息
A new cyclic prefix(CP)-based nonoverlapping FBMC-QAM(CP-NO-FBMC-QAM)system with two prototype filters is proposed in this paper,which satisfies complex orthogonality conditions and good frequency energy confinement at the same *** analyze its inter-carrier interference/inter-symbol interference(ICI/ISI)over multipath *** to the additional CP,the ISI of received symbols over multipath channels is eliminated in the proposed system,and the resulting improvement in the signal-to-interference ratio(SIR)performance is evaluated by theoretical ***,for the ICI caused by multipath propagation in received symbols,we develop a method that eliminates the ICI by frequency-domain channel estimation and equalization before the receiver filtering *** proposed CP-NO-FBMC-QAM system and ICI cancellation method(ICICM)are validated by comparisons of implementation complexity,power spectral density(PSD),bit error rate(BER)and channel estimation performance with conventional CP-based orthogonal frequency division multiplexing(CP-OFDM)and FBMC-QAM systems.
As Internet ofThings(IoT)technologies continue to evolve at an unprecedented pace,intelligent big data control and information systems have become critical enablers for organizational digital transformation,facilitati...
详细信息
As Internet ofThings(IoT)technologies continue to evolve at an unprecedented pace,intelligent big data control and information systems have become critical enablers for organizational digital transformation,facilitating data-driven decision making,fostering innovation ecosystems,and maintaining operational *** this study,we propose an advanced deployment algorithm for Service Function Chaining(SFC)that leverages an enhanced Practical Byzantine Fault Tolerance(PBFT)*** main goal is to tackle the issues of security and resource efficiency in SFC implementation across diverse network *** integrating blockchain technology and Deep Reinforcement Learning(DRL),our algorithm not only optimizes resource utilization and quality of service but also ensures robust security during SFC ***,the enhanced PBFT consensus mechanism(VRPBFT)significantly reduces consensus latency and improves Byzantine node detection through the introduction of a Verifiable Random Function(VRF)and a node reputation grading *** results demonstrate that compared to traditional PBFT,the proposed VRPBFT algorithm reduces consensus latency by approximately 30%and decreases the proportion of Byzantine nodes by 40%after 100 rounds of ***,the DRL-based SFC deployment algorithm(SDRL)exhibits rapid convergence during training,with improvements in long-term average revenue,request acceptance rate,and revenue/cost ratio of 17%,14.49%,and 20.35%,respectively,over existing ***,the CPU resource utilization of the SDRL algorithmreaches up to 42%,which is 27.96%higher than other *** findings indicate that the proposed algorithm substantially enhances resource utilization efficiency,service quality,and security in SFC deployment.
As an important part of electric vehicles, handlebar installation defects are an important factor in the failure of electric vehicles. In the production process, many large-scale manufacturers still assemble the handl...
详细信息
Petri Nets(PNs)are used for modeling and analyzing discreteevent systems,such as communication protocols,trafficsystems,human-computer interaction,and fault ***’state space explosion problem means that the state spac...
详细信息
Petri Nets(PNs)are used for modeling and analyzing discreteevent systems,such as communication protocols,trafficsystems,human-computer interaction,and fault ***’state space explosion problem means that the state spaceof PNs grows exponentially with PNs’*** thefundamental reachability problem is still an NP-Hard problemin *** has been proved that the equivalence problem forthe reachability sets of arbitrary PNs is undecidable except forsome subclass of PNs[1].That is,the reachability problem ofarbitrary PNs cannot be solved ***,there is noefficient and accurate algorithm to solve the problem.10172In recent years,with the emergence of big data and thedevelopment of computing hardware,a series ofbreakthroughs have been achieved in machine learning,suchas AlphaGo,AlphaFold,and ChatGPT[2−4].As a data-drivenapproach,machine learning can learn potential mappingrelationships between inputs and outputs from large-scaledata.
To solve polynomial systems,Harrow,Hassidim,and Lloyd(HHL)proposed a quantum algorithm called HHL *** on the HHL algorithm,Chen et *** an algorithm,the solving the Boolean solutions of polynomial systems(PoSSoB)***,Di...
详细信息
To solve polynomial systems,Harrow,Hassidim,and Lloyd(HHL)proposed a quantum algorithm called HHL *** on the HHL algorithm,Chen et *** an algorithm,the solving the Boolean solutions of polynomial systems(PoSSoB)***,Ding et *** the Boolean Macaulay matrix and analyzed the lower bound on the condition *** by Ding et al.’s research,several related algorithms are proposed in this ***,the improved PoSSoB algorithm using the Boolean Macaulay matrix is proved to have lower ***,for solving equations with errors,a quantum algorithm for the max-polynomial system solving(Max-PoSSo)problem is proposed based on the improved PoSSoB ***,the Max-PoSSo algorithm is extended to the learning with errors(LWE)problem and its special case,the learning parity with noise(LPN)problem,providing a quantitative criterion,the condition number,for the security of these basic problems.
暂无评论