检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Ye, Zhe Mao, Terui Dong, Li Yan, Diqun Faculty of Electrical Engineering and Computer Science Ningbo University Ningbo China Zhejiang Key Laboratory of Mobile Network Application Technology China Ningbo City College of Vocational Technology Ningbo China

Deep speech classification has achieved tremendous success and greatly promoted the emergence of many real-world applications. However, backdoor attacks present a new security threat to it, particularly with untrustworthy third-party platforms, as pre-defined triggers set by the attacker can activate the backdoor. Most of the triggers in existing speech backdoor attacks are sample-agnostic, and even if the triggers are designed to be unnoticeable, they can still be audible. This work explores a backdoor attack that utilizes sample-specific triggers based on voice conversion. Specifically, we adopt a pre-trained voice conversion model to generate the trigger, ensuring that the poisoned samples does not introduce any additional audible noise. Extensive experiments on two speech classification tasks demonstrate the effectiveness of our attack. Furthermore, we analyzed the specific scenarios that activated the proposed backdoor and verified its resistance against fine-tuning. Copyright © 2023, The Authors. All rights reserved.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

A Pseudo-Dual Self-Rectification Framework for Semantic Segmentation

A Pseudo-Dual Self-Rectification Framework for Semantic Segm...

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Huazheng Hao Hui Xiao Li Dong Diqun Yan Dongtai Liang Jiayan Zhuang Chengbin Peng Ningbo University Ningbo China Ningbo Institute of Materials Technology & Engineering Chinese Academy of Sciences Ningbo China Key Laboratory of Mobile Network Application Technology of Zhejiang Province Ningbo China

Semantic segmentation has achieved remarkable success in various applications. However, the training process for such techniques necessitates a significant amount of labeled data. Although semi-supervised frameworks can alleviate this issue, traditional approaches typically require multiple baseline models to form a dual model. To allow a semi-supervised semantic segmentation framework to be used in robotic systems with precious computation and memory resources, we propose a framework utilizing a single baseline model only. The overall framework is composed of three parts: an encoder, a shallow decoder, and a deep decoder. It distills knowledge from the ensemble of two decoders to improve the encoder, which can implicitly form a pseudo-dual model. It also calculates class-wise likelihoods according to the similarity between features and class prototypes learned from different decoders and rectifies low-confidence pseudo-labels. Our framework outperforms state-of-the-art frameworks on benchmark datasets with a significant amount of decrease in using computing resources.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Adversarial defense based on distribution transfer

arXiv

引用

arXiv 2023年

作者： Chen, Jiahao Yan, Diqun Dong, Li Faculty of Electrical Engineering and Computer Science Ningbo University Zhejiang Ningbo315211 China College of Computer Science and Technology Zhejiang University Zhejiang Hangzhou310027 China Key Laboratory of Mobile Network Application Technology of Zhejiang Province Zhejiang Ningbo315211 China

The presence of adversarial examples poses a significant threat to deep learning models and their applications. Existing defense methods provide certain resilience against adversarial examples, but often suffer from decreased accuracy and generalization performance, making it challenging to achieve a trade-off between robustness and generalization. To address this, our paper interprets the adversarial example problem from the perspective of sample distribution and proposes a defense method based on distribution shift, leveraging the distribution transfer capability of a diffusion model for adversarial defense. The core idea is to exploit the discrepancy between normal and adversarial sample distributions to achieve adversarial defense using a pretrained diffusion model. Specifically, an adversarial sample undergoes a forward diffusion process, moving away from the source distribution, followed by a reverse process guided by the protected model (victim model) output to map it back to the normal distribution. Experimental evaluations on CIFAR10 and ImageNet30 datasets are conducted, comparing with adversarial training and input preprocessing methods. For infinite-norm attacks with 8/255 perturbation, accuracy rates of 78.1% and 83.5% are achieved, respectively. For 2-norm attacks with 128/255 perturbation, accuracy rates are 74.3% and 82.5%. Additional experiments considering perturbation amplitude, diffusion iterations, and adaptive attacks also validate the effectiveness of the proposed method. Results demonstrate that even when the attacker has knowledge of the defense, the proposed distribution-based method effectively withstands adversarial examples. It fills the gaps of traditional approaches, restoring high-quality original samples and showcasing superior performance in model robustness and generalization. Copyright © 2023, The Authors. All rights reserved.

关键词： Normal distribution

来源：评论

学校读者我要写书评

暂无评论

Tensor Based Sparse Estimation for RIS-Aided Massive MIMO 23

Tensor Based Sparse Estimation for RIS-Aided Massive MIMO

引用

23rd IEEE International Conference on Communication technology, ICCT 2023

作者： Li, Weilai Wei, Li Huang, Chongwen Yang, Jun Yuen, Chau Zhang, Zhaoyang School of Electrical and Computer Engineering Georgia Institute of Technology Atlanta United States Xidian University State Key Laboratory of Integrated Service Networks Xi'an710071 China School of Electrical and Electronics Engineering Nanyang Technological University Singapore College of Information Science and Electronic Engineering Zhejiang University Hangzhou China Zte Corporation State Key Laboratory of Mobile Network and Mobile Multimedia Technology Shenzhen China

ISBN: (纸本)9798350325959

Reconfigurable intelligent surface (RIS) is intended as a potential technique for future 6-th Generation wireless communication system. However, acquiring accurate channel state information is challenging since the channel matrices are high-dimensional due to large number of reflective elements (REs) in RIS. In this paper, we investigate the problem of sparse channel estimation (CE) in massive multiple-input multiple-output (MIMO) system equipped with large RIS (IRIS), where there are only few REs that are active during CE phase. We propose a two-stage tensor based CE algorithm that exploits the merits of IRIS. Specifically, IRIS act as reflector in the first stage of CE. Then, we identify the condition for the IRIS to activate few specified REs in the second stage. Simulation results demonstrate that the proposed algorithm outperforms existing CE schemes. © 2023 IEEE.

关键词： massive MIMO RIS Sparse channel estimation tensor decomposition

来源：评论

学校读者我要写书评

暂无评论

Gradient Sign Inversion: Making an Adversarial Attack a Good Defense

Gradient Sign Inversion: Making an Adversarial Attack a Good...

引用

International Joint Conference on Neural networks (IJCNN)

作者： Xiaojian Ji Li Dong Rangding Wang Diqun Yan Yang Yin Jinyu Tian Department of Computer Science Ningbo University Zhejiang Key Laboratory of Mobile Network Application Technology Geely Auto Central Research Institute School of Computer Science and Engineering Macau University of Science and Technology

Deep neural networks have been proven vulnerable to deliberately crafted adversarial example, which cause serious safety and security concerns. Many defense approaches were proposed to resist such threats. However, existing defenses such as pre-compression or adversarial training would degrade the model performance on clean images or incur heavy computational costs. In this work, we propose a plug-and-play defensive module Gradient Sign Inversion (GSI) to defend gradient-based attack. Essentially, GSI attempts to inverse the direction of the backpropagated gradient for the victim model, disturbing the adversarial example generation of the attacking while retaining the performance of the vanilla network on genuine inputs. Specifically, an additive model based on periodic trigonometric function is established by investigating the necessary conditions that a suitable defensive module should have. By enforcing constraints on the defensive module, the parameters of GSI are determined, accompanied by a theoretical justification. Interestingly, we observe that the proposed GSI not only prevents the gradient-based adversarial attack, but can even improve the confidence of the ground-truth label when initiating an attack, making the attack betray as a defense. Source code is publicly available at https://***/JidaDiao/GSI.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Towards Ship License Plate Recognition in the Wild: A Large Benchmark and Strong Baseline 39

Towards Ship License Plate Recognition in the Wild: A Large ...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Liu, Baolong Yang, Ruiqing Huang, Roukai Xu, Wenhao Pan, Xin Li, Chuanhuang Wang, Bin Wang, Xun Dong, Jianfeng Zhejiang Gongshang University China Key Laboratory of Public Security Information Application Based on Big-Data Architecture Ministry of Public Security China Zhejiang Key Laboratory of Big Data and Future E-Commerce Technology China Zhejiang University China Network and Data Security China

ISBN: (纸本)157735897X

The paper targets the challenging task of Ship License Plate (SLP) recognition. Existing methods for SLP recognition are hampered by the scarcity of large and publicly available datasets, leading to evaluations on small and non-representative datasets. To alleviate it, we have built a large dataset, called SLP34K, which consists of 34,385 images collected by an intelligent traffic surveillance system. The dataset is carefully manually annotated with text labels and attributes, and presents high data diversity by multiple installation locations and long capturing period of the cameras. Additionally, we propose a simple yet effective SLP recognition baseline method. The baseline is equipped with a strong visual encoder that benefits from initial pre-training via self-supervised learning, followed by further refinement through our devised semantic enhancement module. Extensive experiments on SLP34K verify the effectiveness of our proposed baseline. Moreover, while our baseline is designed for SLP recognition, it can also be used for common scene text recognition and achieve state-of-the-art performance on seven mainstream scene text recognition datasets. © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Three-Dimensional Spatial-Temporal Near-Field Passive Localization Based on an Exact Spatial Propagation Model

Three-Dimensional Spatial-Temporal Near-Field Passive Locali...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Jiaxiong Fang Juan Liu Hua Chen Wei Liu Ye Tian Gang Wang Faculty of Electrical Engineering and Computer Science Ningbo University Ningbo China Zhejiang Key Laboratory of Mobile Network Application Technology Ningbo China School of Electronic Engineering and Computer Science Queen Mary University of London London UK

Based on the exact source-sensor spatial geometry, a three-dimensional (3-D) spatial-temporal localization algorithm for multiple near-field (NF) sources is proposed without adopting the Fresnel approximation, which simplifies the spatial phase difference by Taylors polynomial. In addition, considering the propagation attenuation which varies from different sensors, the spatial and temporal information can be exploited to construct a third-order parallel factor (PARAFAC) data model and the array manifold matrices can be extracted by trilinear decomposition; then, estimation of the unambiguous range and angle parameters of the NF sources is achieved from the spatial amplitude-phase factors by the least squares method. The obtained 3-D parameters associated with each source require no additional pairing process, as also demonstrated by simulation results.

关键词：

来源：评论

学校读者我要写书评

暂无评论

LMCRA: A Reliable Satellite Internet Routing Method with Low Maintenance Cost 17th

LMCRA: A Reliable Satellite Internet Routing Method with Lo...

引用

17th China Conference on Wireless Sensor networks, CWSN 2023

作者： Zhou, Yihu Chen, Haiming Dou, Zhibin Faculty of Electrical Engineering and Computer Science Ningbo University Zhejiang Ningbo315211 China Zhejiang Key Laboratory of Mobile Network Application Ningbo315211 China The 54th Research Institute of China Electronics Technology Group Corporation Shijiazhuang050081 China

ISBN: (纸本)9789819710096

The current shortest path-based satellite routing algorithms do not consider the periodicity and persistence of links, resulting in high costs and unreliable maintenance of routing tables. In this regard, we propose a novel approach that leverages the periodicity and predictability of satellite constellation movements to rid repetitive route calculations and associated maintenance overheads across periods. Furthermore, within each period, we design a routing algorithm, named LMCRA (A Low Maintenance Cost and Reliable Routing Algorithm), which takes into account both the persistence and latency of inter-satellite links. Compared to traditional shortest path algorithms, LMCRA reduces the frequency of path switching caused by neighboring satellites moving out of communication range, thereby enhancing path stability, reducing maintenance costs, and maintaining low path latency. Additionally, a fast rerouting mechanism is considered to enhance the reliability of the routing algorithm. Simulation results demonstrate that, across various network configurations, time spans, and constellation densities, LMCRA significantly improves path stability compared to the conventional Dijkstra algorithm, while maintaining low latency costs. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Routing algorithms

来源：评论

学校读者我要写书评

暂无评论

From Universe to Metaverse: IRS-Assisted Efficient Transmission for Hybrid Earth-Moon network 100

From Universe to Metaverse: IRS-Assisted Efficient Transmiss...

引用

100th IEEE Vehicular technology Conference, VTC 2024-Fall

作者： Lv, Chengcheng Shen, Fei Yan, Feng Shen, Lianfeng Wu, Yi Tang, Liang Zhao, Yu Bu, Zhiyong Chinese Academy of Sciences Key Laboratory of Wireless Sensor Network and Communications Shanghai Institute of Microsystem and Information Technology Shanghai200050 China University of Chinese Academy of Sciences Beijing100049 China Southeast University National Mobile Communications Research Laboratory Nanjing210096 China Fujian Normal University Fujian Provincial Engineering Technology Research Center of Photoelectric Sensing Application Fuzhou350007 China Shanghai Spacecom Satellite Technology Ltd. Shanghai201600 China

ISBN: (纸本)9798331517786

To fulfill the requirements of human lunar exploration programs and establish bases on the moon in the distant future, lunar sensors (LSs) will inevitably produce a significant amount of data. It is necessary to construct the Earth-Moon metaverse in order to obtain and utilize lunar information more effectively. Due to the long communication distance between Earth and Moon as well as the lack of communication resources, a hybrid Earth-Moon metaverse network with channel model and transmission model is designed. To ensure the efficiency and stability of transmission in the network, LSs transmit lunar data to Earth clients (ECs) through the active and passive intelligent reflecting surface (IRS) deployed at relay satellites. Then, we propose a Stackelberg game model to describe the adversarial relationship between the satellites, LSs and ECs, and optimal strategies are obtained by solving the Nash equilibrium to maximize their utility. Simulations demonstrate that the network can effectively shorten the transmission delay and improve the utility of ECs and satellites. © 2024 IEEE.

关键词： Lunar missions

来源：评论

学校读者我要写书评

暂无评论

Mcotm: Mobility-Aware Computation Offloading and Task Migration for Edge Computing in Industrial Iot

SSRN

引用

SSRN 2023年

作者： Qin, Wei Chen, Haiming Wang, Lei Xia, Yinshui Nascita, Alfredo Pescapè, Antonio Zhejiang Provincial Key Laboratory of Mobile Network Application Technology Ningbo University Zhejiang315211 China The Department of Electrical Engineering and Information Technology University of Napoli Federico II Napoli80125 Italy CSEF University of Naples Federico II Italy

Mobility-aware devices are crucial components of Industrial Internet of Things (IIoT). However, they face limitations in terms of battery capacity and computation power, which restrict their ability to provide services requiring broad bandwidth and strong computation power for computation-intensive tasks. While offloading can strengthen device computation power, ineffective offloading decisions result from device mobility and limited adaptability to changes in environmental resources, or are not applicable to the current mobile edge computing (MEC) environment. In this paper, we address these challenges by proposing a mobility-aware computation offloading and task migration approach (MCOTM) based on trajectory and resource prediction to address this issue of mobility offloading, which minimizes task turnaround time and system energy consumption. Simultaneously, our approach enhances the decision agent continuously to decrease task migration rates. MCOTM uses Lagrange interpolation equations to determine the trajectory of mobile devices, and Long Short-Term Memory (LSTM) to track the time-varying resources characteristics in IIoT. These prediction results will be used to assist Deep Deterministic Policy Gradient (DDPG) for making online computation offloading, task migration and resource allocation decisions. Experimental results show that the proposed MCOTM effectively reduces task turnaround time and system energy consumption while maintaining a low task migration rate, even with an increasing number of tasks. © 2023, The Authors. All rights reserved.

关键词： Internet of things

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：