检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Zheng, Hongling Shen, Li Tang, Anke Luo, Yong Hu, Han Du, Bo Wen, Yonggang Tao, Dacheng School of Computer Science School of Artificial Intelligence National Engineering Research Center for Multimedia Software Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University China School of Cyber Science and Technology Shenzhen Campus of Sun Yat-sen University JD Explore Academy China Beijing Institute of Technology China Nanyang Technological University Singapore

Foundation models (FM) have demonstrated remarkable performance across a wide range of tasks (especially in the fields of natural language processing and computer vision), primarily attributed to their ability to comprehend instructions and access extensive, high-quality data. This not only showcases their current effectiveness but also sets a promising trajectory towards the development of artificial general intelligence. Unfortunately, due to multiple constraints, the raw data of the model used for large model training are often inaccessible, so the use of end-to-end models for downstream tasks has become a new research trend, which we call Learn From Model (LFM) in this article. LFM focuses on the research, modification, and design of FM based on the model interface, so as to better understand the model structure and weights (in a black box environment), and to generalize the model to downstream tasks. The study of LFM techniques can be broadly categorized into five major areas: model tuning, model distillation, model reuse, meta learning and model editing. Each category encompasses a repertoire of methods and strategies that aim to enhance the capabilities and performance of FM. This paper gives a comprehensive review of the current methods based on FM from the perspective of LFM, in order to help readers better understand the current research status and ideas. To conclude, we summarize the survey by highlighting several critical areas for future exploration and addressing open issues that require further attention from the research community. The relevant papers we investigated in this article can be accessed at https://***/ruthless-man/Awesome-Learn-from-Model. © 2023, CC BY.

关键词： Frequency modulation

来源：评论

学校读者我要写书评

暂无评论

Federated Learning with Only Positive Labels by Exploring Label Correlations

arXiv

引用

arXiv 2024年

作者： An, Xuming Wang, Dui Shen, Li Luo, Yong Hu, Han Du, Bo Wen, Yonggang Tao, Dacheng School of Information and Electronics Beijing Institute of Technology Beijing100081 China National Engineering Research Center for Multimedia Software School of Computer Science Institute of Artificial Intelligence Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University Wuhan430072 China JD Explore Academy Beijing100000 China School of Computer Science and Engineering Nanyang Technological University Singapore639798 Singapore

Federated learning aims to collaboratively learn a model by using the data from multiple users under privacy constraints. In this paper, we study the multi-label classification problem under the federated learning setting, where trivial solution and extremely poor performance may be obtained, especially when only positive data w.r.t. a single class label are provided for each client. This issue can be addressed by adding a specially designed regularizer on the server-side. Although effective sometimes, the label correlations are simply ignored and thus sub-optimal performance may be obtained. Besides, it is expensive and unsafe to exchange user’s private embeddings between server and clients frequently, especially when training model in the contrastive way. To remedy these drawbacks, we propose a novel and generic method termed Federated Averaging by exploring Label Correlations (FedALC). Specifically, FedALC estimates the label correlations in the class embedding learning for different label pairs and utilizes it to improve the model training. To further improve the safety and also reduce the communication overhead, we propose a variant to learn fixed class embedding for each client, so that the server and clients only need to exchange class embeddings once. Extensive experiments on multiple popular datasets demonstrate that our FedALC can significantly outperform existing counterparts. Copyright © 2024, The Authors. All rights reserved.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

Certificateless Aggregate Signature Scheme with High Efficencicy in Vehicular Ad-hoc network 20

Certificateless Aggregate Signature Scheme with High Efficen...

引用

Proceedings of the 2020 4th International Conference on Electronic Information technology and Computer Engineering

作者： Xiaoming Hu Wenan Tan Chuang Ma Huajie Xu College of Computer and Information Engineering Shanghai Polytechnic University Shanghai China Guangxi Key Laboratory of Multimedia Communications and Network Technology Guangxi University Nanning

ISBN: (纸本)9781450387811

Due to the characteristic of Certificateless aggregate signature (CLASS) scheme, CLASS can be applied into Vehicular Ad-hoc network (VANETs) and other fields. In this paper, we present an improved CLASS scheme for VANETs. In the improved scheme, by using two level dummy identities (one level in the RTA and another level in the RSU), the vehicular can adopt different dummy identity in different RSU or even in the same RSU which can reduce the risk from exposure of one dummy identity and increase the safeguard of the real identity of the vehicular. The improved scheme can protect the real identity of the vehicular. But it is easy to trace the real identity of the vehicular which may break the traffic regulations or do other harmful things. We also give the security analysis of the improved scheme on the unforgeability, blindness and traceability. What's more, the improved scheme has high efficient. It only needs two pairing computation no matter how many signature is aggregated which is more efficient than other similar schemes.

关键词： certificateless signature scheme aggregate signature certificateless aggregate signature scheme network Security VANETs

来源：评论

学校读者我要写书评

暂无评论

Multi-Functional Beamforming Design for Integrated Sensing, communication, and Computation

引用

IEEE Transactions on communications 2024年

作者： Zhao, Yapeng Wu, Qingqing Chen, Wen Zeng, Yong Liu, Ruiqi Mei, Weidong Hou, Fen Ma, Shaodan Shanghai Jiao Tong University Department of Electronic Engineering 200240 China University of Macau State Key Laboratory of Internet of Things for Smart City 999078 China Southeast University National Mobile Communications Research Laboratory Frontiers Science Center for Mobile Information Communication and Security Nanjing210096 China Purple Mountain Laboratories Nanjing211111 China ZTE Corporation State Key Laboratory of Mobile Network and Mobile Multimedia Technology Shenzhen518057 China University of Electronic Science and Technology of China National Key Laboratory of Wireless Communications Chengdu611731 China University of Macau Department of Electrical and Computer Engineering 999078 China

Integrated sensing and communication (ISAC) systems may face a heavy computation burden since the sensory data needs to be further processed. This paper studies a novel system that integrates sensing, communication, and computation, aiming to provide services for different objectives efficiently. This system consists of a multi-antenna multi-functional base station (BS), an edge server, a target, and multiple single-antenna communication users. The BS needs to allocate the available resources to efficiently provide sensing, communication, and computation services. Due to the heavy service burden and limited power budget, the BS can partially offload the tasks to the nearby edge server instead of computing them locally. We consider the estimation of the target response matrix, a general problem in radar sensing, and utilize Cramér-Rao bound (CRB) as the corresponding performance metric. To tackle the non-convex optimization problem, we propose both semidefinite relaxation (SDR)-based alternating optimization and SDR-based successive convex approximation (SCA) algorithms to minimize the CRB of radar sensing while meeting the requirement of communication users and the need for task computing. Furthermore, we demonstrate that the optimal rank-one solutions of both the alternating and SCA algorithms can be directly obtained via the solver or further constructed even when dealing with multiple functionalities. Simulation results show that the proposed algorithms can provide higher target estimation performance than state-of-the-art benchmarks while satisfying the communication and computation constraints. © 1972-2012 IEEE.

关键词： Convex optimization

来源：评论

学校读者我要写书评

暂无评论

Revisiting Token Dropping Strategy in Efficient BERT Pretraining

arXiv

引用

arXiv 2023年

作者： Zhong, Qihuang Ding, Liang Liu, Juhua Liu, Xuebo Zhang, Min Du, Bo Tao, Dacheng 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 Research Center for Graphic Communication Printing and Packaging Institute of Artificial Intelligence Wuhan University China Institute of Computing and Intelligence Harbin Institute of Technology China University of Sydney Australia

Token dropping is a recently-proposed strategy to speed up the pretraining of masked language models, such as BERT, by skipping the computation of a subset of the input tokens at several middle layers. It can effectively reduce the training time without degrading much performance on downstream tasks. However, we empirically find that token dropping is prone to a semantic loss problem and falls short in handling semantic-intense tasks (§2). Motivated by this, we propose a simple yet effective semantic-consistent learning method (SCTD) to improve the token dropping. SCTD aims to encourage the model to learn how to preserve the semantic information in the representation space. Extensive experiments on 12 tasks show that, with the help of our SCTD, token dropping can achieve consistent and significant performance gains across all task types and model sizes. More encouragingly, SCTD saves up to 57% of pretraining time and brings up to +1.56% average improvement over the vanilla token dropping. Copyright © 2023, The Authors. All rights reserved.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

A distributionally robust self-scheduling under price uncertainty based on CVaR

arXiv

引用

arXiv 2021年

作者： Yang, Linfeng Yang, Ying Chen, Guo Dong, Zhaoyang School of Computer Electronics and Information Guangxi University Nanning530004 China Guangxi Key Laboratory of Multimedia Communication and Network Technology Guangxi University School of Electrical Engineering and Tele-communications University of NSW SydneyNSW2052 Australia

To ensure a successful bid while maximizing of profits, generation companies (GENCOs) need a self-scheduling strategy that can cope with a variety of scenarios. So distributionally robust optimization (DRO) is a good choice because that it can provide an adjustable self-scheduling strategy for GENCOs in the uncertain environment, which can well balance robustness and economics compared to strategies derived from robust optimization (RO) and stochastic programming (SO). In this paper, a novel moment-based DRO model with conditional value-at-risk (CVaR) is proposed to solve the self-scheduling problem under electricity price uncertainty. Such DRO models are usually translated into semi-definite programming (SDP) for solution, however, solving large-scale SDP needs a lot of computational time and resources. For this shortcoming, two effective approximate models are proposed: one approximate model based on vector splitting and another based on alternate direction multiplier method (ADMM), both can greatly reduce the calculation time and resources, and the second approximate model only needs the information of the current area in each step of the solution and thus information private is guaranteed. Simulations of three IEEE test systems are conducted to demonstrate the correctness and effectiveness of the proposed DRO model and two approximate models. Copyright © 2021, The Authors. All rights reserved.

关键词： Stochastic programming

来源：评论

学校读者我要写书评

暂无评论

FedABC: Targeting Fair Competition in Personalized Federated Learning

arXiv

引用

arXiv 2023年

作者： Wang, Dui Shen, Li Luo, Yong Hu, Han Su, Kehua Wen, Yonggang Tao, Dacheng National Engineering Research Center for Multimedia Software School of Computer Science Institute of Artificial Intelligence Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University China Hubei Luojia Laboratory Wuhan China JD Explore Academy China School of Information and Electronics Beijing Institute of Technology China School of Computer Science and Engineering Nanyang Technological University Singapore

Federated learning aims to collaboratively train models without accessing their client’s local private data. The data may be Non-IID for different clients and thus resulting in poor performance. Recently, personalized federated learning (PFL) has achieved great success in handling Non-IID data by enforcing regularization in local optimization or improving the model aggregation scheme on the server. However, most of the PFL approaches do not take into account the unfair competition issue caused by the imbalanced data distribution and lack of positive samples for some classes in each client. To address this issue, we propose a novel and generic PFL framework termed Federated Averaging via Binary Classification, dubbed FedABC. In particular, we adopt the "one-vs-all" training strategy in each client to alleviate the unfair competition between classes by constructing a personalized binary classification problem for each class. This may aggravate the class imbalance challenge and thus a novel personalized binary classification loss that incorporates both the under-sampling and hard sample mining strategies is designed. Extensive experiments are conducted on two popular datasets under different settings, and the results demonstrate that our FedABC can significantly outperform the existing counterparts. Copyright © 2023, The Authors. All rights reserved.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Spatial-Temporal Space Hand-in-Hand: Spatial-Temporal Video Super-Resolution via Cycle-Projected Mutual Learning

arXiv

引用

arXiv 2022年

作者： Hu, Mengshun Jiang, Kui Liao, Liang Xiao, Jing Jiang, Junjun Wang, Zheng National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence School of Computer Science Wuhan University China Hubei Key Laboratory of Multimedia and Network Communication Engineering China Nanyang Technological University Singapore Harbin Institute of Technology China

Spatial-Temporal Video Super-Resolution (ST-VSR) aims to generate super-resolved videos with higher resolution (HR) and higher frame rate (HFR). Quite intuitively, pioneering two-stage based methods complete ST-VSR by directly combining two sub-tasks: Spatial Video Super-Resolution (S-VSR) and Temporal Video Super-Resolution (T-VSR) but ignore the reciprocal relations among them. Specifically, 1) T-VSR to S-VSR: temporal correlations help accurate spatial detail representation with more clues;2) S-VSR to T-VSR: abundant spatial information contributes to the refinement of temporal prediction. To this end, we propose a one-stage based Cycle-projected Mutual learning network (CycMu-Net) for ST-VSR, which makes full use of spatial-temporal correlations via the mutual learning between S-VSR and T-VSR. Specifically, we propose to exploit the mutual information among them via iterative up-and-down projections, where the spatial and temporal features are fully fused and distilled, helping the high-quality video reconstruction. Besides extensive experiments on benchmark datasets, we also compare our proposed CycMu-Net with S-VSR and T-VSR tasks, demonstrating that our method significantly outperforms state-of-the-art methods. Codes are publicly available at: https://***/hhhhhumengshun/CycMuNet. Copyright © 2022, The Authors. All rights reserved.

关键词： Optical resolving power

来源：评论

学校读者我要写书评

暂无评论

A DV-Hop positioning algorithm based on the glowworm swarm optimisation of mixed chaotic strategy

引用

International Journal of Security and networks 2019年第1期14卷 23-33页

作者： Song, Ling Zhao, Liqin Ye, Jin School of Computer and Electronic Information Guangxi Key Laboratory of Multimedia Communications and Network Technology Guangxi University Nanning530004 China

DV-Hop, as a typical location algorithm without ranging, is widely used in node localisation of wireless sensor networks. However, in the third phase of DV-Hop, a least square method is used to solve the nonlinear equations. Using this method to locate the unknown nodes will produce large coordinates errors, poor stability of positioning accuracy, low location coverage and high energy consumption. An improved localisation algorithm based on hybrid chaotic strategy (MGDV-Hop) is proposed in this paper. Firstly, a glowworm swarm optimisation of hybrid chaotic strategy based on chaotic mutation and chaotic inertial weight updating (MC-GSO) is proposed. Then, MC-GSO is used to replace the least square method in estimating node coordinates. By establishing the error fitness function, the linear solution of coordinates is transformed into a two-dimensional combinatorial optimisation problem. Simulation results show that the average location error is reduced, while the location coverage is increased and the energy consumption is decreased. © 2019 Inderscience Enterprises Ltd.

关键词： Energy utilization

来源：评论

学校读者我要写书评

暂无评论

Bayesian Framework for Multi-User Cooperative Radio SLAM

Bayesian Framework for Multi-User Cooperative Radio SLAM

引用

IEEE International Symposium on Personal, Indoor and Mobile Radio communications (PIMRC)

作者： Hang Que Jie Yang Tao Du Shuqiang Xia Chao-Kai Wen Shi Jin National Mobile Communications Research Laboratory Southeast University Nanjing China Frontiers Science Center for Mobile Information Communication and Security Southeast University Nanjing China Key Laboratory of Measurement and Control of Complex Systems of Engineering Ministry of Education Southeast University Nanjing China ZTE Corporation Shenzhen China State Key Laboratory of Mobile Network and Mobile Multimedia Technology Shenzhen China Institute of Communications Engineering National Sun Yat-sen University Kaohsiung Taiwan

ISBN: (数字)9798350362244

ISBN: (纸本)9798350362251

The advancement of millimeter-wave communication technology heralds new sensing capabilities. By leveraging channel multipath parameter estimates, we can harness simultaneous localization and mapping (SLAM) for precise user equipment (UE) localization and radio map construction in 6 G communication systems. Particularly in multi-UE scenarios, SLAM empowers base stations to amalgamate the local radio maps of various UEs efficiently. This study introduces a novel Bayesian framework specifically designed for multi-UE SLAM, complemented by a tailored factor graph. We also unveil a two-stage multi-UE SLAM algorithm. Our simulation results reveal that this algorithm substantially enhances radio map construction accuracy by $\mathbf{4 8. 5 \%}$ and UE localization accuracy by $13.5 \%$, outperforming single-UE cases. Moreover, the algorithm demonstrates remarkable adaptability to environmental changes, showcasing its potential for long-term evolution in dynamic settings.

关键词： Location awareness Simultaneous localization and mapping Accuracy communication systems Heuristic algorithms Simulation Bayes methods Sensors Millimeter wave communication Land mobile radio

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：