检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Cao, Hongye Feng, Fan Huo, Jing Yang, Shangdong Fang, Meng Yang, Tianpei Gao, Yang National Key Laboratory for Novel Software Technology Nanjing University Nanjing210093 China Department of Electrical Engineering City University of Hong Kong Hong Kong School of Computer Science Nanjing University of Posts and Telecommunications Nanjing210023 China Department of Computer Science University of Liverpool LiverpoolL69 3BX United Kingdom

Model-based offline Reinforcement Learning (RL) constructs environment models from offline datasets to perform conservative policy optimization. Existing approaches focus on learning state transitions through ensemble models, rollouting conservative estimation to mitigate extrapolation errors. However, the static data makes it challenging to develop a robust policy, and offline agents cannot access the environment to gather new data. To address these challenges, we introduce Model-based Offline Reinforcement learning with AdversariaL data augmentation (MORAL). In MORAL, we replace the fixed horizon rollout by employing adversaria data augmentation to execute alternating sampling with ensemble models to enrich training data. Specifically, this adversarial process dynamically selects ensemble models against policy for biased sampling, mitigating the optimistic estimation of fixed models, thus robustly expanding the training data for policy optimization. Moreover, a differential factor is integrated into the adversarial process for regularization, ensuring error minimization in extrapolations. This data-augmented optimization adapts to diverse offline tasks without rollout horizon tuning, showing remarkable applicability. Extensive experiments on D4RL benchmark demonstrate that MORAL outperforms other model-based offline RL methods in terms of policy learning and sample efficiency. © 2025, CC BY.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

PCA: Semi-Supervised Segmentation with Patch Confidence Adversarial Training

引用

IEEE Transactions on Network science and engineering 2025年

作者： Xu, Zhenghua Yang, Runhe Xu, Zihang Zhang, Shuo Yang, Yuchen Liu, Weipeng Xu, Weichao Chen, Junyang Lukasiewicz, Thomas Leung, Victor C. M. Hebei University of Technology State Key Laboratory of Reliability and Intelligence of Electrical Equipment School of Health Sciences and Biomedical Engineering Tianjin China The University of Hong Kong Department of Electrical and Electronic Engineering Hong Kong The Johns Hopkins University Department of Applied Mathematics and Statistic Baltimore United States Hebei University of Technology School of Artificial Intelligence Tianjin China Hebei Provincial Hospital of Traditional Chinese Medicine Digestive Disease Center Shijiazhuang China Shenzhen University College of Computer Science and Software Engineering Shenzhen China University of Oxford Department of Computer Science University of Oxford United Kingdom Department of Computer Science United Kingdom Vienna University of Technology institute of Logic and Computation nstitute of Logic and Computation Vienna Austria

Deep-learning-based semi-supervised learning (SSL) methods have achieved a strong performance in medical image segmentation, which can alleviate doctors' expensive annotation by utilizing a large amount of unlabeled data. Unlike most existing semi-supervised learning methods, adversarial training methods distinguish samples from different sources by learning the data distribution of the segmentation map, leading the segmenter to generate more accurate predictions. We argue that the current performance restrictions for such approaches are the problems of feature extraction and learning preferences. In this paper, we propose a new semi-supervised adversarial method called Patch Confidence Adversarial Training (PCA) for medical image segmentation. The PCA method's discriminator penalizes patch-level structures, guiding the generator to optimize different patch areas, by leveraging pixel context, the generator is driven to focus on high-frequency features, making it harder to deceive the discriminator and easy to converge to an ideal state, which more effectively guides the segmenter to generate high-quality pseudo-labels. Furthermore, at the discriminator's input, we supplement image information constraints, making it simpler to fit the expected data distribution. Extensive experiments on the Automated Cardiac Diagnosis Challenge (ACDC) 2017 dataset and the Brain Tumor Segmentation (BraTS) 2019 challenge dataset show that our method outperforms the state-of-the-art semi-supervised methods, which demonstrates its effectiveness for medical image segmentation. © 2025 IEEE. All rights reserved.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Task Delay and Energy Consumption Minimization for Low-altitude MEC via Evolutionary Multi-objective Deep Reinforcement Learning

arXiv

引用

arXiv 2025年

作者： Sun, Geng Ma, Weilong Li, Jiahui Sun, Zemin Wang, Jiacheng Niyato, Dusit Mao, Shiwen College of Computer Science and Technology Jilin University Changchun130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun130012 China College of Computing and Data Science Nanyang Technological University Singapore639798 Singapore College of Software Jilin University Changchun130012 China Department of Electrical and Computer Engineering Auburn University AuburnAL36849-5201 United States

The low-altitude economy (LAE), driven by unmanned aerial vehicles (UAVs) and other aircraft, has revolutionized fields such as transportation, agriculture, and environmental monitoring. In the upcoming six-generation (6G) era, UAV-assisted mobile edge computing (MEC) is particularly crucial in challenging environments such as mountainous or disaster-stricken areas. The computation task offloading problem is one of the key issues in UAV-assisted MEC, primarily addressing the trade-off between minimizing the task delay and the energy consumption of the UAV. In this paper, we consider a UAV-assisted MEC system where the UAV carries the edge servers to facilitate task offloading for ground devices (GDs), and formulate a calculation delay and energy consumption multi-objective optimization problem (CDECMOP) to simultaneously improve the performance and reduce the cost of the system. Then, by modeling the formulated problem as a multi-objective Markov decision process (MOMDP), we propose a multi-objective deep reinforcement learning (DRL) algorithm within an evolutionary framework to dynamically adjust the weights and obtain non-dominated policies. Moreover, to ensure stable convergence and improve performance, we incorporate a target distribution learning (TDL) algorithm. Simulation results demonstrate that the proposed algorithm can better balance multiple optimization objectives and obtain superior non-dominated solutions compared to other methods. Copyright © 2025, The Authors. All rights reserved.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Family of Mutually Uncorrelated Codes for DNA Storage Address Design

IEEE Transactions on Nanobioscience

引用

IEEE Transactions on Nanobioscience 2025年 PP卷 PP页

作者： Liu, Zhenlu Cao, Ben Shao, Qi Zheng, Yanfen Wang, Bin Zhou, Shihua Zheng, Pan Dalian University Key Laboratory of Advanced Design and Intelligent Computing Ministry of Education School of Software Engineering Dalian116622 China Dalian University of Technology School of Computer Science and Technology Dalian116024 China University of Canterbury Upper Riccarton Department of Accounting and Information Systems Christchurch8140 New Zealand

Deoxyribonucleic acid (DNA) has become an ideal medium for long-term storage and retrieval due to its extremely high storage density and long-term stability. But access efficiency is an existing bottleneck in DNA storage, especially the lack of high-quality random access address sequences. Therefore, in this paper, we report a series of approaches based on k-weakly mutually uncorrelated (k-WMU) codes to design the address sequence to improve the access efficiency of DNA storage. To address the problem of DNA sequences that are poorly scalable at the base level, we propose a 0-m-ruling coding scheme combined with k-WMU codes that can make address sequences avoid generating secondary structure with stem lengths ranging from 3 to 9. Based on the decoupled structure, We further extend the k-WMU codes with error correction function while satisfying combinatorial biological constraints. In order to investigate the performance of the designed address sequences for real-world applications, we perform simulation experiments based on thermodynamic properties and error correction capability as well as compared the minimum free energy (MFE), melting temperature (TM), and average decoding success rate (ADSR) with previous work. The results show that designed address sequences have a high MFE value and ADSR and a substantial reduction in TM-variance while satisfying the combinatorial biological constraints. As the quality of address sequences improves, this will help to achieve accurate random access as well as enhance the robustness of the DNA storage system. © 2002-2011 IEEE.

关键词： DNA sequences

来源：评论

学校读者我要写书评

暂无评论

HyperDiff: Masked Diffusion Model with High-efficient Transformer for Hyperspectral Image Cross-Scene Classification

HyperDiff: Masked Diffusion Model with High-efficient Transf...

引用

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

作者： Pei Zhang Dong Wang Chanyue Wu Jing Yang Lei Kang Zongwen Bai Ying Li Qiang Shen School of Computer Science Northwestern Polytechnical University Xi’an China Shaanxi Key Laboratory of Intelligent Processing for Big Energy Data Yan’an University Yan’an China School of Automation and Software Engineering Shanxi University Taiyuan China Qingdao Topscomm Communication Co. Ltd Qingdao China Department of Computer Science Aberystwyth University UK

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Hyperspectral Image (HSI) cross-scene classification is a challenging task in remote sensing, particularly when real-time processing of Target Domain (TD) HSI is required, and data cannot be reused for training. While deep learning methods have shown promising results, the generalization ability of HSI representations remains limited, mainly due to class label imbalance. This paper introduces a dual-stage learning framework based on transfer learning to enhance classification accuracy in the TD. The framework includes a self-supervised learning stage and a supervised fine-tuning stage. The self-supervised stage focuses on learning robust representations by leveraging inherent structures within HSI data, while the fine-tuning stage uses training labels to extract semantic information. A masked diffusion model predicts masked tokens from unmasked ones, capturing both high-level structures and fine details in HSI data. An efficient spatiospectral Transformer, which removes self-attention from the decoder, is proposed to enhance the self-supervised process. This design allows mask tokens to obtain information from visible tokens without interacting with each other, reducing sequence length and computational costs. By decoding each mask token conditionally independently, only a subset of masked tokens is processed. Extensive experiments on two public HSI datasets demonstrate that the proposed method outperforms state-of-the-art techniques.

关键词： Training Transfer learning Self-supervised learning Transformers Diffusion models Real-time systems Decoding Computational efficiency Speech processing Hyperspectral imaging

来源：评论

学校读者我要写书评

暂无评论

Generalized Cross-Entropy Benchmarking for Random Circuits with Ergodicity

arXiv

引用

arXiv 2025年

作者： Cheng, Bin Meng, Fei Zhang, Zhi-Jiong Yung, Man-Hong Shenzhen Institute for Quantum Science and Engineering Department of Physics Southern University of Science and Technology Shenzhen518055 China Centre for Quantum Technologies National University of Singapore Singapore Centre for Quantum Software and Information Faculty of Engineering and Information Technology University of Technology SydneyNSW2007 Australia Department of Physics City University of Hong Kong Tat Chee Avenue Kowloon Hong Kong QICI Quantum Information and Computation Initiative Department of Computer Science The University of Hong Kong Pokfulam Road Hong Kong International Quantum Academy Shenzhen518048 China Guangdong Provincial Key Laboratory of Quantum Science and Engineering Southern University of Science and Technology Shenzhen518055 China Shenzhen Key Laboratory of Quantum Science and Engineering Southern University of Science and Technology Shenzhen518055 China

Cross-entropy benchmarking is a central technique used to certify a quantum chip in recent experiments. To better understand its mathematical foundation and develop new benchmarking schemes, we introduce the concept of ergodicity to random circuit sampling and find that the Haar random quantum circuit satisfies an ergodicity condition—the average of certain types of post-processing function over the output bit strings is close to the average over the unitary ensemble. For noiseless random circuits, we prove that the ergodicity holds for polynomials of degree t with positive coefficients and when the random circuits form a unitary 2t-design. For strong enough noise, the ergodicity condition is violated. This suggests that ergodicity is a property that can be exploited to certify a quantum chip. We formulate the deviation of ergodicity as a measure for quantum chip benchmarking and show that it can be used to estimate the circuit fidelity for global depolarizing noise and weakly correlated noise. For a quadratic post-processing function, our framework recovers Google’s result on estimating the circuit fidelity via linear cross-entropy benchmarking (XEB), and we give a sufficient condition on the noise model characterizing when such estimation is valid. Our results establish an interesting connection between ergodicity and noise in random circuits and provide new insights into designing quantum benchmarking schemes. Copyright © 2025, The Authors. All rights reserved.

关键词： Quantum electronics

来源：评论

学校读者我要写书评

暂无评论

Correction to: A comprehensive performance evaluation of container migration strategies

引用

Computing 2025年第4期107卷 1-2页

作者： Feitosa, Leonel Barbosa, Vandirleya Sabino, Arthur Lima, Luiz Nelson Fé, Iure Silva, Luis Guilherme Callou, Gustavo Carvalho, Juliana Leão, Erico Nguyen, Tuan Anh Rego, Paulo Silva, Francisco Airton Laboratory of Applied Research to Distributed Systems (PASID) Federal University of Piauí (UFPI) Picos Brazil Department of Computing Federal Rural University of Pernambuco (UFRPE) Recife Brazil Department of Software Engineering Faculty of Information Technology University of Industry and Trade (HUIT) Ho Chi Minh City Vietnam Department of Computer Science Federal University of Ceará (UFC) Fortaleza Brazil

来源：评论

学校读者我要写书评

暂无评论

Cacomp: A Cloud-assisted Collaborative Deep Learning Compiler Framework for DNN Tasks on Edge

引用

IEEE Transactions on computers 2025年

作者： Lin, Weiwei Lin, Jinhui Zhang, Haotong Wu, Wentai Wu, Weizheng Li, Zhetao Li, Keqin South China University of Technology School of Computer Science and Engineering Guangzhou510641 China Pengcheng Laboratory Shenzhen518066 China South China University of Technology School of Software Engineering Guangzhou510641 China Jinan University Department of Computer Science College of Information Science and Technology Guangzhou510632 China Jinan University College of Information Science and Technology Guangzhou510632 China State University of New York Department of Computer Science New PaltzNY12561 United States

With the development of edge computing, DNN services have been widely deployed on edge devices. The deployment efficiency of deep learning models relies on the optimization of inference and scheduling policy. However, traditional optimization methods on edge devices still suffer from prohibitively long tuning time due to devices’ low computational power. Meanwhile, the widely used scheduling algorithm, the dominant resource fairness algorithm(DRF algorithm), struggles to maximize the efficiency of model execution on edge devices and inevitably increases average waiting time as it is not applicable in the real-time distributed computing environment. In this paper, we propose Cacomp, a distributed cloud-assisted deep learning compiler framework that features accelerating the optimization on edge devices with assistance from the cloud and a novel inference task scheduling algorithm. Our framework utilizes the tuning records from the cloud devices and proposes a two-step distillation strategy to obtain the best tuning record set for the edge device. For the scheduling process, we propose an RD-DRF algorithm to allocate inference tasks to edge devices based on dominant resource matching in real time. Extensive results show that our framework can achieve up to 2.19× improvement in the optimization time compared with other methods on edge devices. Our proposed scheduling algorithm significantly shortens the average waiting time of inference tasks by 30% and improves resource utilization by 20% on edge devices. © 1968-2012 IEEE.

关键词： Distributed cloud

来源：评论

学校读者我要写书评

暂无评论

GPT-Based Automated Induction: Vulnerability Detection in Medical software

引用

IEEE Journal of Biomedical and Health Informatics 2025年 PP卷 PP页

作者： Deng, Liangjun Lei, Hang Khan, Fazlullah Srivastava, Gautam Chen, Jingxue Haque, Mainul University of Electronic Science and Technology of China School of Information and Software Engineering Chengdu610054 China University of Nottingham Ningbo China School of Computer Science and the Next-Generation Internet of Everything Laboratory Faculty of Science and Engineering Zhejiang Ningbo315104 China Lovely Professional University Division of Research and Development Phagwara India Chandigarh University University Centre of Research and Development Mohali India Brandon University Department of Math and Computer Science BrandonMBR7A 6A9 Canada China Medical University Research Centre for Interneural Computing Taichung40402 Taiwan Chitkara University Institute of Engineering and Technology Chitkara University Centre for Research Impact & Outcome Punjab Rajpura140401 India University of Nottingham Ningbo China Department of Mathematics Faculty of Science and Engineering Zhejiang Ningbo315104 China

Integrating Natural Language Processing (NLP) with Generative Pre-trained Transformer (GPT) models plays a pivotal role in enhancing the accuracy and efficiency of healthcare software, which is essential for patient safety and providing high-quality care. The precision of healthcare software is fundamental to protecting the well-being of the patient. In addition, it can ensure the delivery of superior care, maintain the integrity of healthcare systems, and promote trust and cost-effectiveness. It is necessary to emphasize the importance of software reliability in its development and deployment. Symbolic execution serves as a vital technology in automated vulnerability detection. However, symbolic execution often faces problems such as path explosion, which seriously affects efficiency. Although there have been several studies to reduce the number of computational paths in symbolic execution, this problem remains a major obstacle. Therefore, more efficient solutions are urgently needed to ensure the software security. This paper proposes a large-scale language model(LLM) induction method mitigating path explosion applied to symbolic execution engines. In contrast to traditional symbolic execution engines, which often result in timeout or out-of-memory detection, our approach achieves the task of detecting vulnerabilities in seconds. Furthermore, our proposal improves the scalability of symbolic execution, allowing more extensive and complex programs to be analyzed without significant increases in computational resources or time. This scalability is crucial to tackling modern software systems and improving the efficiency and effectiveness of automated defect verification in healthcare software. © 2025 IEEE. All rights reserved,

关键词： Scalability

来源：评论

学校读者我要写书评

暂无评论

Continual Test-Time Adaptation for Single Image Defocus Deblurring via Causal Siamese Networks

arXiv

引用

arXiv 2025年

作者： Cui, Shuang Li, Yi Li, Jiangmeng Tang, Xiongxin Su, Bing Xu, Fanjiang Xiong, Hui National Key Laboratory of Space Integrated Information System Institute of Software Chinese Academy of Sciences University of Chinese Academy of Sciences China Beijing Key Laboratory of Big Data Management and Analysis Methods Gaoling School of Artificial Intelligence Renmin University of China China Department of Computer Science & Engineering the Hong Kong University of Science and Technology Hong Kong

Single image defocus deblurring (SIDD) aims to restore an all-in-focus image from a defocused one. Distribution shifts in defocused images generally lead to performance degradation of existing methods during out-of-distribution inferences. In this work, we gauge the intrinsic reason behind the performance degradation, which is identified as the heterogeneity of lens-specific point spread functions. empirical evidence supports this finding, motivating us to employ a continual test-time adaptation (CTTA) paradigm for SIDD. However, traditional CTTA methods, which primarily rely on entropy minimization, cannot sufficiently explore task-dependent information for pixel-level regression tasks like SIDD. To address this issue, we propose a novel Siamese networks-based continual test-time adaptation framework, which adapts source models to continuously changing target domains only requiring unlabeled target data in an online manner. To further mitigate semantically erroneous textures introduced by source SIDD models under severe degradation, we revisit the learning paradigm through a structural causal model and propose Causal Siamese networks (CauSiam). Our method leverages large-scale pre-trained vision-language models to derive discriminative universal semantic priors and integrates these priors into Siamese networks, ensuring causal identifiability between blurry inputs and restored images. Extensive experiments demonstrate that CauSiam effectively improves the generalization performance of existing SIDD methods in continuously changing domains. Copyright © 2025, 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

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：