检索结果-南通市图书馆

FMSA:a meta-learning framework-based fast model stealing attack technique against intelligent network intrusion detection systems

在线全文

学校读者我要写书评

暂无评论

Cybersecurity 2024年第1期7卷 110-121页

作者： Kaisheng Fan Weizhe zhang Guangrui Liu Hui He School of Cyberspace Science Harbin Institute of TechnologyHarbin 150001China Department of New Networks Peng Cheng LaboratoryShenzhen 518055GuangdongChina

Intrusion detection systems are increasingly using machine *** machine learning has shown excellent performance in identifying malicious traffic,it may increase the risk of privacy *** paper focuses on imple-menting a model stealing attack on intrusion detection *** model stealing attacks are hard to imple-ment in practical network environments,as they either need private data of the victim dataset or frequent access to the victim *** this paper,we propose a novel solution called Fast Model Stealing Attack(FMSA)to address the problem in the field of model stealing *** also highlight the risks of using ML-NIDS in network ***,meta-learning frameworks are introduced into the model stealing algorithm to clone the victim model in a black-box ***,the number of accesses to the target model is used as an optimization term,resulting in minimal queries to achieve model ***,adversarial training is used to simulate the data distribution of the target model and achieve the recovery of privacy *** experiments on multiple public datasets,compared to existing state-of-the-art algorithms,FMSA reduces the number of accesses to the target model and improves the accuracy of the clone model on the test dataset to 88.9%and the similarity with the target model to 90.1%.We can demonstrate the successful execution of model stealing attacks on the ML-NIDS system even with protective measures in place to limit the number of anomalous queries.

关键词： AI security Model stealing attack Network intrusion detection Meta learning

Parametric instability in the pure-quartic nonlinear Schrödinger equation

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Physics B 2024年第3期33卷 316-321页

作者：张云红刘冲 School of Physics Northwest UniversityXi'an 710127China Shaanxi Key Laboratory for Theoretical Physics Frontiers Xi'an 710127China peng Huanwu Center for Fundamental Theory Xi'an 710127China

We study the nonlinear stage of modulation instability(MI)in the non-intergrable pure-quartic nonlinear Schrödinger equation where the fourth-order dispersion is modulated *** the three-mode truncation,we reveal the complex recurrence of parametric resonance(PR)breathers,where each recurrence is associated with two oscillation periods(PR period and internal oscillation period).The nonlinear stage of parametric instability admits the maximum energy exchange between the spectrum sidebands and central mode occurring outside the MI gain band.

关键词： modulation instability parametric resonance breather three-mode truncation energy exchange

Understanding adversarial attacks on observations in deep reinforcement learning

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2024年第5期67卷 69-83页

作者： You QIAOBEN Chengyang YING Xinning ZHOU Hang SU Jun ZHU Bo zhang Department of Computer Science and Technology Beijing National Research Center for Information Science and Technology Tsinghua-Bosch Joint Center for Machine Learning Institute for Artificial Intelligence Tsinghua University peng Cheng Laboratory

Deep reinforcement learning models are vulnerable to adversarial attacks that can decrease the cumulative expected reward of a victim by manipulating its observations. Despite the efficiency of previous optimization-based methods for generating adversarial noise in supervised learning, such methods might not achieve the lowest cumulative reward since they do not generally explore the environmental ***, a framework is provided to better understand the existing methods by reformulating the problem of adversarial attacks on reinforcement learning in the function space. The reformulation approach adopted herein generates an optimal adversary in the function space of targeted attacks, repelling them via a generic two-stage framework. In the first stage, a deceptive policy is trained by hacking the environment and discovering a set of trajectories routing to the lowest reward or the worst-case performance. Next, the adversary misleads the victim to imitate the deceptive policy by perturbing the observations. Compared to existing approaches, it is theoretically shown that our adversary is strong under an appropriate noise level. Extensive experiments demonstrate the superiority of the proposed method in terms of efficiency and effectiveness, achieving state-of-the-art performance in both Atari and MuJoCo environments.

关键词： deep learning reinforcement learning adversarial robustness adversarial attack

Dual-wavelength pumped latticed Fermi-Pasta-Ulam recurrences in nonlinear Schrödinger equation

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Chinese Physics B 2024年第3期33卷 277-280页

作者：张倩姚献坤董恒 School of Physics Northwest UniversityXi'an 710127China Shaanxi Key Laboratory for Theoretical Physics Frontiers Xi'an 710127China peng Huanwu Center for Fundamental Theory Xi'an 710127China

We show that the nonlinear stage of the dual-wavelength pumped modulation instability(MI)in nonlinear Schrödinger equation(NLSE)can be effectively analyzed by mode truncation *** resulting complicated heteroclinic structure of instability unveils all possible dynamic trajectories of nonlinear ***,the latticed-Fermi-Pasta-Ulam recurrences on the modulated-wave background in NLSE are also investigated and their dynamic trajectories run along the Hamiltonian contours of the heteroclinic *** is demonstrated that there has much richer dynamic behavior,in contrast to the nonlinear waves reported *** novel nonlinear wave promises to inject new vitality into the study of MI.

关键词： modulation instability dual-wavelength pumps latticed-Fermi-Pasta-Ulam recurrences

Learning topological defects formation with neural networks in a quantum phase transition

在线全文

学校读者我要写书评

暂无评论

Communications in Theoretical Physics 2024年第5期76卷 68-76页

作者： Han-Qing Shi Hai-Qing zhang Center for Gravitational Physics Department of Space ScienceBeihang UniversityBeijing 100191China peng Huanwu Collaborative Center for Research and Education Beihang UniversityBeijing 100191China

Neural networks possess formidable representational power,rendering them invaluable in solving complex quantum many-body *** they excel at analyzing static solutions,nonequilibrium processes,including critical dynamics during a quantum phase transition,pose a greater challenge for neural *** address this,we utilize neural networks and machine learning algorithms to investigate time evolutions,universal statistics,and correlations of topological defects in a one-dimensional transverse-field quantum Ising ***,our analysis involves computing the energy of the system during a quantum phase transition following a linear quench of the transverse magnetic field *** excitation energies satisfy a power-law relation to the quench rate,indicating a proportional relationship between the excitation energy and the kink ***,we establish a universal power-law relationship between the first three cumulants of the kink numbers and the quench rate,indicating a binomial distribution of the ***,the normalized kink-kink correlations are also investigated and it is found that the numerical values are consistent with the analytic formula.

关键词： neural networks machine learning transverse-field quantum Ising model kibble-zurek mechanism

Rethinking multi-spatial information for transferable adversarial attacks on speaker recognition systems

在线全文

学校读者我要写书评

暂无评论

CAAI Transactions on Intelligence Technology 2024年第3期9卷 620-631页

作者： Junjian zhang Hao Tan Le Wang Yaguan Qian Zhaoquan Gu Guangzhou University GuangzhouChina Department of New Networks Peng Cheng LaboratoryShenzhenChina Harbin Institute of Technology(Shenzhen) ShenzhenChina Zhejiang University of Science and Technology HangzhouChina

Adversarial attacks have been posing significant security concerns to intelligent systems,such as speaker recognition systems(SRSs).Most attacks assume the neural networks in the systems are known beforehand,while black-box attacks are proposed without such information to meet practical *** black-box attacks improve trans-ferability by integrating multiple models or training on multiple datasets,but these methods are *** by the optimisation strategy with spatial information on the perturbed paths and samples,we propose a Dual Spatial Momentum Iterative Fast Gradient Sign Method(DS-MI-FGSM)to improve the transferability of black-box at-tacks against ***,DS-MI-FGSM only needs a single data and one model as the input;by extending to the data and model neighbouring spaces,it generates adver-sarial examples against the integrating *** reduce the risk of overfitting,DS-MI-FGSM also introduces gradient masking to improve *** authors conduct extensive experiments regarding the speaker recognition task,and the results demonstrate the effectiveness of their method,which can achieve up to 92%attack success rate on the victim model in black-box scenarios with only one known model.

关键词： speaker recognition spoofing attacks

Intellicise Model Transmission for Semantic Communication in Intelligence-Native 6G Networks

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

China Communications 2024年第7期21卷 95-112页

作者： Wang Yining Han Shujun Xu xiaodong Meng Rui Liang Haotai Dong Chen zhang Ping State Key Laboratory of Network and Switching Technology Beijing University of Posts and TelecommunicationsBeijing 100876China Department of Broadband Communication Peng Cheng LaboratoryShenzhen 518066China

To facilitate emerging applications and demands of edge intelligence(EI)-empowered 6G networks,model-driven semantic communications have been proposed to reduce transmission volume by deploying artificial intelligence(AI)models that provide abilities of semantic extraction and ***,it is not feasible to preload all AI models on resource-constrained ***,in-time model transmission becomes a crucial *** paper proposes an intellicise model transmission architecture to guarantee the reliable transmission of models for semantic *** mathematical relationship between model size and performance is formulated by employing a recognition error function supported with experimental *** consider the characteristics of wireless channels and derive the closed-form expression of model transmission outage probability(MTOP)over the Rayleigh ***,we define the effective model accuracy(EMA)to evaluate the model transmission performance of both communication and *** we propose a joint model selection and resource allocation(JMSRA)algorithm to maximize the average EMA of all *** results demonstrate that the average EMA of the JMSRA algorithm outperforms baseline algorithms by about 22%.

关键词： edge intelligence(EI) model transmission outage probability and accuracy resource allocation semantic communication

Transmissive reconfigurable metasurface enabling independent control of active and passive modules through weak coupling

在线全文

学校读者我要写书评

暂无评论

Photonics Research 2024年第7期12卷 1449-1456页

作者： KUN XUE HENG WEI CILEI zhang YONGHAO zhang HAOLIANG SUN SHAOHUA DONG peng Cheng Laboratory Shenzhen 518055China Department of Electrical and Computer Engineering National University of SingaporeSingapore 117583Singapore State Key Laboratory of Radio Frequency Heterogeneous Integration College of Electronics and Information EngineeringShenzhen UniversityShenzhen 518060China

Metasurfaces have demonstrated rich electromagnetic control capabilities and degrees of freedom in past years. As is well known, for passive metasurfaces, their functionalities cannot be further expanded accordingly once prototypes are established. Therefore, reconfigurable metasurfaces, utilizing active devices to replace geometric changes in passive structures, have received widespread attention, especially with the development of wireless communication recently. In reconfigurable metasurfaces, artificial meta-atoms are composed of active devices and passive structures combined together. However, these two modules are usually utilized as a whole due to the tight coupling of the active devices and the passive structures, which results in passive structures not receiving sufficient attention and being utilized as independent degrees of freedom. In this article, we propose the concept of weakly coupled reconfigurable metasurfaces in transmissive systems, enabling independent control of active and passive modules through weak coupling. As the proof of concept, a simple weakly coupled system is proposed, which can realize the transmission wavefront engineering through the geometric changes of meta-structures in passive mode, while achieving switching between transmission and reflection states in active mode, respectively. Our exploration lies in making use of the physical structure, which is easily neglected in traditional reconfigurable metasurface design, emphasizing the collaborative work of active and passive modules, exploring more available variables within the same aperture, and providing a potential solution for balancing functionality and resource consumption in practical applications.

关键词： modules coupling weakly

High-speed GaN-based laser diode with modulation bandwidth exceeding 5 GHz for 20 Gbps visible light communication

在线全文

学校读者我要写书评

暂无评论

Photonics Research 2024年第6期12卷 1186-1193页

作者： JUNFEI WANG JUNHUI HU CHAOWEN GUAN YUQI HOU ZENGYI XU LEIHAO SUN YUE WANG YUNING ZHOU BOON S.OOI JIANYANG SHI ZIWEI LI JUNWEN zhang NAN CHI SHAOHUA YU CHAO SHEN School of Information Science and Technology Fudan UniversityShanghai 200433China peng Cheng Laboratory Shenzhen 518000China ZGC Institute of Ubiquitous-X Innovation and Applications Beijing 100876China Photonics Laboratory King Abdullah University of Science and Technology(KAUST)Thuwal 23955-6900Saudi Arabia

Visible light communication(VLC)based on laser diodes demonstrates great potential for high data rate maritime,terrestrial,and aerial wireless data ***,we design and fabricate high-speed blue laser diodes(LDs)grown on c-plane gallium nitride(GaN)*** was achieved through active region design and miniaturization toward a narrow ridge waveguide,short cavity length,and single longitudinal mode Fabry–Perot laser *** fabricated mini-LD has a low threshold current of 31 mA and slope efficiency of 1.02 W/A.A record modulation bandwidth of 5.9 GHz(−3dB)was measured from the *** the developed mini-LD as a transmitter,the VLC link exhibits a high data transmission rate of 20.06 Gbps adopting the bit and power loading discrete multitone(DMT)modulation *** corresponding bit error rate is 0.003,satisfying the forward error correction *** demonstrated GaN-based mini-LD has significantly enhanced data transmission rates,paving the path for energy-efficient VLC systems and integrated photonics in the visible regime.

关键词： communication diode waveguide