检索结果-内蒙古大学图书馆

Securely Fine-tuning Pre-trained Encoders Against Adversarial Examples

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhou, Ziqi Li, Minghui Liu, Wei Hu, Shengshan Zhang, Yechao Wan, Wei Xue, Lulu Zhang, Leo Yu Yao, Dezhong Jin, Hai National Engineering Research Center for Big Data Technology and System China Services Computing Technology and System Lab. China Cluster and Grid Computing Lab. China Hubei Engineering Research Center on Big Data Security China Hubei Key Laboratory of Distributed System Security China School of Computer Science and Technology Huazhong University of Science and Technology China School of Software Engineering Huazhong University of Science and Technology China School of Cyber Science and Engineering Huazhong University of Science and Technology China School of Information and Communication Technology Griffith University Australia

With the evolution of self-supervised learning, the pre-training paradigm has emerged as a predominant solution within the deep learning landscape. Model providers furnish pre-trained encoders designed to function as versatile feature extractors, enabling downstream users to harness the benefits of expansive models with minimal effort through fine-tuning. Nevertheless, recent works have exposed a vulnerability in pre-trained encoders, highlighting their susceptibility to downstream-agnostic adversarial examples (DAEs) meticulously crafted by attackers. The lingering question pertains to the feasibility of fortifying the robustness of downstream models against DAEs, particularly in scenarios where the pre-trained encoders are publicly accessible to the attackers. In this paper, we initially delve into existing defensive mechanisms against adversarial examples within the pre-training paradigm. Our findings reveal that the failure of current defenses stems from the domain shift between pre-training data and downstream tasks, as well as the sensitivity of encoder parameters. In response to these challenges, we propose Genetic Evolution-Nurtured Adversarial Fine-tuning (Gen-AF), a two-stage adversarial fine-tuning approach aimed at enhancing the robustness of downstream models. Gen-AF employs a genetic-directed dual-track adversarial fine-tuning strategy in its first stage to effectively inherit the pre-trained encoder. This involves optimizing the pre-trained encoder and classifier separately while incorporating genetic regularization to preserve the model’s topology. In the second stage, Gen-AF assesses the robust sensitivity of each layer and creates a dictionary, based on which the top-k robust redundant layers are selected with the remaining layers held fixed. Upon this foundation, we conduct evolutionary adaptability fine-tuning to further enhance the model’s generalizability. Our extensive experiments, conducted across ten self-supervised training methods and six

关键词： Topology

DRL-M4MR: An Intelligent Multicast Routing Approach Based on DQN Deep Reinforcement Learning in SDN

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhao, Chenwei Ye, Miao Xue, Xingsi Lv, Jianhui Jiang, Qiuxiang Wang, Yong School of Information and Communication Guilin University of Electronic Technology Guilin541004 China Guangxi Key Laboratory of Wireless Wideband Communication and Signal Processing Guilin University of Electronic Technology Guilin541004 China Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology Fujian Fuzhou350118 China Peng Cheng Lab. Guangdong Shenzhen518038 China School of Computer Science and Information Security Guilin University of Electronic Technology Guilin541004 China

Traditional multicast routing methods have some problems in constructing a multicast tree, such as limited access to network state information, poor adaptability to dynamic and complex changes in the network, and inflexible data forwarding. To address these defects, the optimal multicast routing problem in software-defined networking (SDN) is tailored as a multi-objective optimization problem, and an intelligent multicast routing algorithm DRL-M4MR based on the deep Q network (DQN) deep reinforcement learning (DRL) method is designed to construct a multicast tree in SDN. First, the multicast tree state matrix, link bandwidth matrix, link delay matrix, and link packet loss rate matrix are designed as the state space of the DRL agent by combining the global view and control of the SDN. Second, the action space of the agent is all the links in the network, and the action selection strategy is designed to add the links to the current multicast tree under four cases. Third, single-step and final reward function forms are designed to guide the intelligence to make decisions to construct the optimal multicast tree. The experimental results show that, compared with existing algorithms, the multicast tree construct by DRL-M4MR can obtain better bandwidth, delay, and packet loss rate performance after training, and it can make more intelligent multicast routing decisions in a dynamic network environment. Copyright © 2022, The Authors. All rights reserved.

关键词： Reinforcement learning

Finite-Blocklength Multi-Antenna Covert communication Aided By A UAV Relay

学校读者我要写书评

暂无评论

Finite-Blocklength Multi-Antenna Covert Communication Aided ...

2021 IEEE Global communications Conference (GLOBECOM)

作者： Xinying Chen Min Sheng Nan Zhao Wei Xu Dusit Niyato School of Information and Communication Engineering Dalian University of Technology Dalian China Faculty of Information Technology University of Jyväskylä Jyväskylä Finland State Key Laboratory of Integrated Services Networks Xidian University Xi'an China National Mobile Communications Research Lab. Southeast University Nanjing Jiangsu China School of Computer Science and Engineering Nanyang Technological University Singapore Singapore

ISBN: (纸本)9781728181059

We propose a UAV-relayed covert communication scheme with finite blocklength to maximize the effective transmission rate from the transmitter to the legitimate receiver against a flying warden. The transmitter adopts the maximum ratio transmission and the relay performs Gaussian signaling transmission to cause uncertainty at the warden. First, the optimal detection thresholds are derived at the warden towards the transmitter and the relay, respectively. Then, the hovering location of the warden is optimized to maximize the summation of relative entropies from the transmitter and the relay, which can greatly threaten the covertness. With this worst covert situation, the blocklength and transmit power at the transmitter and the relay are jointly optimized under the constraint of the error detection probability to maximize the effective transmission rate from the transmitter to the legitimate receiver. Numerical results are provided to demonstrate the effectiveness of the proposed UAV-relayed covert communication scheme.

关键词： Uncertainty Transmitters Conferences Receivers Autonomous aerial vehicles Entropy Global communication

Cooperative Reflection and Synchronization Design for Distributed Multiple-RIS communications

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhao, Yaqiong Xu, Wei You, Xiaohu Wang, Ning Sun, Huan The National Mobile Communications Research Lab. Frontiers Science Center for Mobile Information Communication and Security Southeast University Nanjing210096 China Purple Mountain Laboratories Nanjing211111 China The School of Information Engineering Zhengzhou University Zhengzhou450001 China The Wireless Technology Laboratory Huawei Technologies Co. Ltd. Shanghai201206 China

To reap the promised gain achieved by distributed reconfigurable intelligent surface (RIS)-enhanced communications in a wireless network, timing synchronization among these metasurfaces is an essential prerequisite in practice. This paper proposes a unified framework for the joint estimation of the unknown timing offsets and the RIS channel parameters, as well as the design of cooperative reflection and synchronization algorithm for the distributed multiple-RIS communication. Considering that RIS is usually a passive device with limited capability of signal processing, the individual timing offset and channel gains of each hop of the RIS links cannot be directly estimated. To make the estimation tractable, we propose to estimate the cascaded channels and timing offsets jointly by deriving a maximum likelihood estimator. Furthermore, we theoretically characterize the Cramér-Rao lower bound (CRLB) to evaluate the accuracy of this estimator. By using the proposed estimator and the derived CRLBs, an efficient resynchronization algorithm is devised jointly at the RISs and the destination to compensate the multiple timing offsets. Based on the majorization-minimization framework, the proposed algorithm admits semi-closed and closed form solutions for the RIS reflection matrices and the timing offset equalizer, respectively. Simulation results verify that our theoretical analysis well matches the numerical tests and validate the effectiveness of the proposed resynchronization algorithm. Copyright © 2022, The Authors. All rights reserved.

关键词： Timing circuits

Intelligent reflecting surface aided MISO uplink communication network: Feasibility and power minimization for perfect and imperfect CSI

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Liu, Yang Zhao, Jun Li, Ming Wu, Qingqing School of Information and Communication Engineering Dalian University of Technology Dalian China School of Computer Science and Engineering Nanyang Technological University Singapore National Mobile Communications Research Laboratory Southeast University Nanjing210096 China State Key Lab. of Internet of Things for Smart City University of Macau Macau China

In this paper, we consider the weighted sum-power minimization under quality-of-service (QoS) constraints in the multi-user multi-input-single-output (MISO) uplink wireless network assisted by intelligent reflecting surface (IRS). We perform a comprehensive investigation on various aspects of this problem. First, when users have sufficient transmit powers, we present a new sufficient condition guaranteeing arbitrary information rate constraints. This result strengthens the feasibility condition in existing literature. Then, we design novel penalty dual decomposition (PDD) based and nonlinear equality constrained alternative direction method of multipliers (neADMM) based solutions to tackle the IRS-dependent-QoS-constraints, which effectively solve the feasibility check and power minimization problems. Besides, we further extend our proposals to the cases where channel status information (CSI) is imperfect and develop an online stochastic algorithm that satisfy QoS constraints stochastically without requiring prior knowledge of CSI errors. Extensive numerical results are presented to verify the effectiveness of our proposed algorithms. © 2020, CC BY.

关键词： Quality of service

Capacity and Outage Analysis of Sm-Otfs System with Imperfect Csi in V2v communications

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Bai, Zhiquan Wang, Runlai Yang, Yingchao Hu, Huili Li, Jingxin Zhou, Xiao Wang, Chengyou Dai, Jian Shandong Provincial Key Lab. of Wireless Communication Technologies School of Information Science and Engineering Shandong University Qingdao266237 China School of Basic Science Shandong University Jinan250012 China School of Mechanical Electrical and Information Engineering Shandong University Weihai264209 China School of Mechatronical Engineering Beijing Institute of Technology Beijing100081 China Key Laboratory for Experimental Teratology of Ministry of Education Department of Physiology

With the continuous emergence of high-mobility communication scenarios, such as the Internet of Vehicles (IoV) and Vehicle to Vehicle (V2V) communications, more unique challenges have appeared in mobile communications, due to the severe Doppler frequency shift and fast time-varying channel caused by high mobility. Meanwhile, the moving speed, data volume, and quality of service are becoming more and more important in V2V communications. Providing efficient and reliable wireless communication services to high-mobility users has become a critical issue. Spatial modulation (SM) based orthogonal time frequency space (OTFS) (SM-OTFS) system can improve the reliability and effectiveness of V2V communications because of the excellent Doppler shift resistance of OTFS modulation and the low complexity of SM transmission. In this paper, considering the case that achieving perfect channel estimation is really challenging in actual situation, we derive and analyze the capacity and outage performance of the SM-OTFS system under the circumstance of ideal pulse and imperfect channel state information (CSI) based on the statistical probability and the delay-Doppler domain (DD) input-output relationship. Our theoretical analysis and derivation are approved by the numerical results. Moreover, we also demonstrate the effect of the number of resolvable multipaths, the error of channel estimation, and the different moving speeds on the performance of the SM-OTFS system in V2V communications. © 2024, The Authors. All rights reserved.

关键词： Channel estimation

Robust Generalization against Photon-Limited Corruptions via Worst-Case Sharpness Minimization

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Huang, Zhuo Zhu, Miaoxi Xia, Xiaobo Shen, Li Yu, Jun Gong, Chen Han, Bo Du, Bo Liu, Tongliang Syndey Ai Centre The University of Sydney Australia National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence School of Computer Science Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University China Jd Explore Academy China Department of Automation University of Science and Technology of China China Key Lab. of Intelligent Percept. and Syst. for High-Dimensional Information of Ministry of Education School of Computer Science and Engineering Nanjing University of Science and Technology China Department of Computer Science Hong Kong Baptist University Hong Kong

Robust generalization aims to tackle the most challenging data distributions which are rare in the training set and contain severe noises, i.e., photon-limited corruptions. Common solutions such as distributionally robust optimization (DRO) focus on the worst-case empirical risk to ensure low training error on the uncommon noisy distributions. However, due to the over-parameterized model being optimized on scarce worst-case data, DRO fails to produce a smooth loss landscape, thus struggling on generalizing well to the test set. Therefore, instead of focusing on the worst-case risk minimization, we propose SharpDRO by penalizing the sharpness of the worst-case distribution, which measures the loss changes around the neighbor of learning parameters. Through worst-case sharpness minimization, the proposed method successfully produces a flat loss curve on the corrupted distributions, thus achieving robust generalization. Moreover, by considering whether the distribution annotation is availab.e, we apply SharpDRO to two problem settings and design a worst-case selection process for robust generalization. Theoretically, we show that SharpDRO has a great convergence guarantee. Experimentally, we simulate photon-limited corruptions using CIFAR10/100 and ImageNet30 datasets and show that SharpDRO exhibits a strong generalization ability against severe corruptions and exceeds well-known baseline methods with large performance gains. Copyright © 2023, The Authors. All rights reserved.

关键词： Optimization

Efficient Channel Estimation for Millimeter Wave and Terahertz Systems Enabled by Integrated Super-resolution Sensing and communication

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xu, Jingran Wang, Huizhi Zeng, Yong Xu, Xiaoli Wu, Qingqing Yang, Fei Chen, Yan Jamalipour, Abbas The National Mobile Communications Research Laboratory Frontiers Science Center for Mobile Information Communication and Security Southeast University Nanjing210096 China The Purple Mountain Laboratories Nanjing211111 China Department of Electronic Engineering Shanghai Jiao Tong University Shanghai200240 China The Wireless Technology Lab. 2012 Lab Shanghai Huawei Technologies Co. Ltd. Shanghai201206 China The School of Electrical and Computer Engineering The University of Sydney CamperdownNSW2006 Australia

Integrated super-resolution sensing and communication (ISSAC) has emerged as a promising technology to achieve extremely high precision sensing for those key parameters, such as the angles of the sensing targets. In this paper, we propose an efficient channel estimation scheme enabled by ISSAC for millimeter wave (mmWave) and TeraHertz (THz) systems with a hybrid analog/digital beamforming architecture, where both the pilot overhead and the cost of radio frequency (RF) chains are significantly reduced. The key idea is to exploit the fact that subspace-based super-resolution algorithms such as multiple signal classification (MUSIC) can estimate channel parameters accurately without requiring dedicate a priori known pilots. In particular, the proposed method consists of two stages. First, the angles of the multi-path channel components are estimated in a pilot-free manner during the transmission of data symbols. Second, the multi-path channel coefficients are estimated with very few pilots. Compared to conventional channel estimation schemes that rely solely on channel training, our approach requires the estimation of much fewer parameters in the second stage. Furthermore, with channel multi-path angles obtained, the beamforming gain can be achieved when pilots are sent to estimate the channel path gains. To comprehensively investigate the performance of the proposed scheme, we consider both the basic line-of-sight (LoS) channels and more general multipath channels. We compare the performance of the minimum mean square error (MMSE) of channel estimation and the resulting beamforming gains of our proposed scheme with the traditional scheme that rely exclusively on channel training. It is demonstrated that our proposed method significantly outperforms the benchmarking scheme. Simulation results are presented to validate our theoretical findings. Copyright © 2024, The Authors. All rights reserved.

关键词： Mean square error

Reports of the AAAI 2010 fall symposia

学校读者我要写书评

暂无评论

Reports of the AAAI 2010 fall symposia

作者： Azevedo, Roger Biswas, Gautam Bohus, Dan Carmichael, Ted Finlayson, Mark A. Hadzikadic, Mirsad Havasi, Catherine Horvitz, Eric Kanda, Takayuki Koyejo, Oluwasanmi Lawless, William F. Lenat, Doug Meneguzzi, Felipe Mutlu, Bilge Oh, Jean Pirrone, Roberto Raux, Antoine Sofge, Donald A. Sukthankar, Gita Van Durme, Benjamin Yorke-Smith, Neil Department of Educational and Counseling Psychology McGill University Canada Center for Intelligent Systems Vanderbilt University United States Microsoft Research Redmond United States Department of Software and Information Systems University of North Carolina Charlotte United States Complex Systems Institute United States Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology United States Media Lab. Massachusetts Institute of Technology United States ATR Intelligent Robotics and Communication Laboratories Kyoto Japan Department of Electrical and Computer Engineering University of Texas Austin United States Department of Mathematics Paine College Augusta GA United States Cycorp Austin TX United States Robotics Institute Carnegie Mellon University United States Department of Computer Sciences University of Wisconsin Madison United States University of Palermo Italy Honda Research Institute United States Navy Center for Applied Research in Artificial Intelligence Naval Research Laboratory Washington DC United States Department of Electrical Engineering and Computer Science University of Central Florida United States Department of Computer Science Johns Hopkins University United States American University of Beirut Lebanon SRI International's Artificial Intelligence Center United States

The Association for the Advancement of Artificial Intelligence (AAAI) presented the 2010 Fall Symposium Series on November 11-13, 2010. The eight symposia included Cognitive and Metacognitive Educational Systems, Commonsense Knowledge, Complex Adaptive Systems: Resilience, Robustness, and Evolvability, Computational Models of Narrative, Dialog with Robots, Manifold Learning and Its Applications, Proactive Assistant Agents and Quantum Informatics for Cognitive, Social, and Semantic Processes. Cognitive and Metacognitive Educational Systems aimed to provide a comprehensive definition of metacognitive educational systems that is inclusive of the theoretical, architectural, and educational aspects of this field. The AAAI Commonsense Knowledge Fall Symposium had the goal of bringing together the diverse elements of this community whose work benefits from or contributes to the representation of general knowledge about the world. One of the specific goals of Proactive symposium was to gather the researchers from various projects in assistant agents to share their wisdom in retrospect.

关键词： Semantics