检索结果-内蒙古大学图书馆

Unsupervised AoA Estimation Based on Dual-Path Knowledge-Aware Auto-Encoders

IEEE Transactions on Wireless Communications 2025年第5期24卷 4425-4439页

作者： Guo, Zhiheng Xiao, Yuanzhang Wang, Xijun Chen, Xiang University of Hawaii Hawaii Advanced Wireless Technologies Institute HonoluluHI United States Sun Yat-sen University School of Electronics and Information Technology Guangzhou510275 China The Chinese University of Hong Kong Hong Kong Guangdong Provincial Key Laboratory of Big Data Computing Shenzhen China

In this paper, an unsupervised deep learning-based framework based on dual-path model-driven auto-encoders (AE) is proposed for angle-of-arrivals (AoAs) estimation in massive MIMO systems. Specifically designed for AoA estimation, the proposed framework differs from the conventional AE in two aspects. Firstly, unlike conventional auto-encoders, our framework employs a dual-path neural network for the encoder, decoupling the estimated parameters and enabling independent updates of each paths. Secondly, the decoder has fixed weights that implement the signal propagation model, instead of learnable parameters. This knowledge-aware decoder ensures the output of meaningful physical parameters (i.e., AoAs) which is unattainable by conventional AEs. We also conduct a thorough analysis to characterize the multiple global optima and local optima of the estimation problem. This analysis inspires the design of a low-complexity two-phase training scheme and confirms the convergence of our proposed framework. Consequently, our framework addresses two key challenges in unsupervised learning: the lack of interpretability and the convergence to local optima. Extensive simulations validate our theoretical analysis and demonstrate the performance improvements of our proposed framework. © 2002-2012 IEEE.

关键词： Unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

Network intrusion detection based on feature fusion of attack dimension

引用

Journal of Supercomputing 2025年第6期81卷 1-33页

作者： Sun, Xiaolong Gu, Zhengyao Zhang, Hao Gu, Jason Liu, Yanhua Dong, Chen Ye, Junwei College of Computer and Data Science Fuzhou University Fuzhou350116 China Fujian Key Laboratory of Network Computing and Intelligent Information Processing Fuzhou University Fuzhou350116 China Engineering Research Center of Big Data Intelligence Chinese Ministry of Education Fuzhou350116 China Department of Electrical and Computer Engineering Dalhousie University HalifaxNSB3J 1Z1 Canada

Network traffic anomaly detection involves the rapid identification of intrusions within a network through the detection, analysis, and classification of network traffic data. The variety of cyberattacks encompasses diverse attack principles. Employing an indiscriminate feature selection strategy may lead to the neglect of key features highly correlated with specific attack types. This oversight could diminish the recognition rate for that category, thereby impacting the overall performance of the detection model. To address this issue, this paper proposes a network traffic anomaly detection model based on the fusion of attack-dimensional features. Firstly, construct binary classification datasets independently for each attack class and perform individual feature selection to extract positively correlated features for each class. The features are then fused by employing a combination methods. Subsequently, based on the fused sub-datasets, base classifiers are trained. Finally, an ensemble learning approach is introduced to integrate the predictions of individual classifiers, enhancing the robustness of the model. The proposed approach, validated on NSL-KDD and UNSW-NB15 benchmark datasets, outperforms the latest methods in the field by achieving a 2% and 7% increase in precision on weighted averages. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025.

关键词： Feature Selection

来源：评论

学校读者我要写书评

暂无评论

Controlled-source electromagnetic noise attenuation via a deep convolutional neural network and high-quality sounding curve screening mechanism

引用

Geophysics 2025年第3期90卷 WA125-WA140页

作者： Liu, Yecheng Li, Diquan Li, Jin Zhang, Xian Central South University Monitoring Ministry of Education Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Changsha China Hunan Provincial Key Laboratory of Non-ferrous Resources and Geological Hazard Detection Changsha China Central South University School of Geoscience and Info-physics Changsha China Hunan Normal University College of Information Science and Engineering Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing Changsha China Hunan University of Finance and Economics School of Information Technology and Management Hunan Provincial Key Laboratory of Finance & Economics Big Data Science and Technology Changsha China

Strong noise is one of the biggest challenges in controlled-source electromagnetic (CSEM) exploration, which severely affects the quality of the recorded signal. We develop a novel and effective CSEM noise attenuation method that integrates a deep convolutional neural network (DCNN) with a high-quality sounding curve screening mechanism. This method first uses a DCNN to learn the large-scale noise characteristics of CSEM measured data and build a DCNN denoising model for attenuating the noise. Because some impulse noise remains after DCNN denoising, the high-quality sounding curve screening mechanism is innovatively applied to optimize the DCNN denoised data. The core of this method is the smoothness value (SV), which quantitatively evaluates the smoothness of the sounding curve (apparent resistivity curve). In effect, the mechanism segments the overall DCNN denoised data and calculates the corresponding sounding curve and its SV for each segment. The highest-quality sounding curve can be screened out by comparing the SVs of all curves as well as obtaining the highest-quality data segment present in the overall DCNN denoised data. Finally, our method is validated with synthetic and field data, demonstrating its feasibility and effectiveness in attenuating large-scale CSEM noise. The quality of CSEM data, which is heavily affected by noise, is significantly improved after processing by our method. © 2025 Society of Exploration Geophysicists. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

DMC-Watermark: A backdoor richer watermark for dual identity verification by dynamic mask covering

引用

Applied Intelligence 2025年第8期55卷 1-15页

作者： Zhu, Yujia Wang, Ruoxi Xia, Daoxun School of Big Data and Computer Science Guizhou Normal University Guiyang China Guizhou Key Laboratory of Advanced Computing Guizhou Normal University Guiyang China Engineering Laboratory for Applied Technology of Big Data in Education Guizhou Normal University Guiyang China

With the increasing use of neural networks, the importance of copyright protection for these models has gained significant attention. Backdoor watermarking is one of the key methods for protecting copyright. However, on the one hand, most existing backdoor watermarks are triggered by visual images, making them easily detectable, and therefore vulnerable to various attacks. On the other hand, it is difficult for these methods to carry information related to the creator’s identity which can easily lead to fraudulent claims of ownership. These factors contribute to the vulnerability and limitations of backdoor watermarking. In this paper, we propose DMC-Watermark, a backdoor richer watermarking method that uses dynamic mask-covered image structures as triggers. Leveraging the semantic preservation of image structure in transformation attacks, we select image structure as triggers. Furthermore, we convert the author-related information into an array of color information and apply it as a mask to the extracted image structures, allowing it to serve as a second layer of verification during the validation phase to resist fraudulent claims of ownership. The final trigger pattern, embedded with author-related image structures, is applied to the selected images in the trigger set, generating a final trigger set that is trained together with clean samples to produce a protected model. The experiments show that the proposed DMC-Watermark performs well in terms of fidelity, invisibility, undetectability, functionality, dual verification and robustness on three different datasets and four representative DNNs, and it has wide applicability and excellent results in high-resolution images.

关键词： Backdoor watermark DNN copyright Dual verification Image structure extraction Model protection

来源：评论

学校读者我要写书评

暂无评论

ApriDS: An evidence linkage network construction model for judgment documents

引用

Journal of King Saud University - Computer and Information Sciences 2025年第1期37卷 1-13页

作者： Zhou, Yulin Qin, Yongbin Lin, Chuan Text Computing & Cognitive Intelligence Engineering Research Center of National Education Ministry Guizhou University Guiyang 550025 China State Key Laboratory of Public Big Data Guizhou University Guiyang 550025 China College of Computer Science & Technology Guizhou University Guiyang 550025 China

Evidence serves as the basis for determining facts in the judicial trial process, and exploring the correlation between evidence has become an essential task. However, there is uncertainty and unreliability of evidence. In this paper, we propose an ApriDS model that can mine correlations between evidence and construct evidence linkage networks. Specifically, we use the association rule algorithm to mining for the correlations between evidence in order to initially construct an evidence linkage network. Moreover, we also combine the imprecise reasoning theory to alleviate unreliability among evidence. Finally, combining the above effectively mitigates uncertainty and unreliability in the evidence linkage network. We conduct extensive experiments and present analyses on five case datasets, including marriages, thefts, homicides, contract disputes and unlawful detention. Compared to the baseline model, our method’s accuracy improves by 12.75%, 14.16%, 14.80%, 21.64% and 26.15% on the five datasets, respectively. © The Author(s) 2025.

关键词： Apriori algorithm D-S evidence theory Evidence linkage network Judgment document

来源：评论

学校读者我要写书评

暂无评论

Extraction and Area Calculation of Chicken Comb and Wattle Based on YOLO v7 and Optimized U-Net Network

引用

Nongye Jixie Xuebao/Transactions of the Chinese Society for Agricultural Machinery 2025年第4期56卷 415-426页

作者： Yang, Duanli Shen, Hongshuo Chen, Hui Gao, Yuan College of Information Science and Technology Hebei Agricultural University Baoding071001 China Hebei Key Laboratory of Agricultural Big Data Baoding071001 China College of Animal Science and Technology Hebei Agricultural University Baoding071001 China Key Laboratory of Broiler and Layer Facilities Engineering Ministry of Agriculture and Rural Affairs Baoding071001 China

Traditional manual measurement of poultry comb and wattle areas poses contact-induced stress risks, zoonotic disease transmission hazards, and substantial measurement errors. A non-contact measurement system integrating YOLO v7 with an improved U - Net architecture was proposed. Three key innovations were presented: a dual-stage detection framework utilizing YOLO v7 for head pose screening and ROI extraction, effectively eliminating off-angle image interference;a novel CoT - UNet model incorporating Contextual Transformer blocks into U - Net's encoder for dynamic-static context fusion, combined with DyC - UP module employing dynamically adjustable convolution kernels to enhance irregular edge detection;a pixel-area conversion algorithm achieving precise spatial mapping through calibration coefficients. Experimental results demonstrated significant improvements;the enhanced CoT-UNet outperformed baseline models by 4. 77 percentage points (comb) and 8. 75 percentage points (wattle) in loll, along with 5.31 percentage points and 5.06 percentage points precision gains respectively. Absolute measurement errors for comb (0. 62 ∼ 3. 50 cm2) and wattle (0. 10 ∼ 2. 93 cm2) showed marked superiority over manual methods (3.58 ∼ 7.27 cm2). Multi-scenario validation revealed stable relative errors of 2.41% ∼ 13.62% for combs and 1.00% ∼ 29.21% for wattles across varied postures (three types), angles (two positions), and distances (two levels). This automated system enabled stress-free poultry biometric measurement, providing reliable technical support for intelligent breeding selection. © 2025 Chinese Society of Agricultural Machinery. All rights reserved.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

IMPRESS: an importance-informed multi-tier prefix KV storage system for large language model inference 25

IMPRESS: an importance-informed multi-tier prefix KV storage...

引用

Proceedings of the 23rd USENIX Conference on File and Storage Technologies

作者： Weijian Chen Shuibing He Haoyang Qu Ruidong Zhang Siling Yang Ping Chen Yi Zheng Baoxing Huai Gang Chen The State Key Laboratory of Blockchain and Data Security Zhejiang University and Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security and Zhejiang Key Laboratory of Big Data Intelligent Computing The State Key Laboratory of Blockchain and Data Security Zhejiang University Huawei Cloud

ISBN: (纸本)9781939133458

Modern advanced large language model (LLM) applications often prepend long contexts before user queries to improve model output quality. These contexts frequently repeat, either partially or fully, across multiple queries. Existing systems typically store and reuse the keys and values of these contexts (referred to as prefix KVs) to reduce redundant computation and time to first token (TTFT). When prefix KVs need to be stored on disks due to insufficient CPU memory, reusing them does not always reduce TTFT, as disk I/O latency is high. In this paper, we propose IMPRESS, an importance-informed multi-tier prefix KV storage system to reduce I/O delay for LLM inference by only loading important prefix KVs. IMPRESS first leverages the insight that there is significant similarity in important token index sets across attention heads and introduces an I/O-efficient important KV identification algorithm. It then optimizes prefix KV storage and caching through importance-informed KV management, reducing TTFT during model inference. Our experimental results show that IMPRESS can reduce TTFT by up to 2.8× compared to state-of-the-art systems, while maintaining comparable inference accuracy.

关键词：

来源：评论

学校读者我要写书评

暂无评论

OmniThink: Expanding Knowledge Boundaries in Machine Writing through Thinking

arXiv

引用

arXiv 2025年

作者： Xi, Zekun Yin, Wenbiao Fang, Jizhan Wu, Jialong Fang, Runnan Zhang, Ningyu Jiang, Yong Xie, Pengjun Huang, Fei Chen, Huajun Zhejiang University China Tongyi Lab Alibaba Group China Zhejiang Key Laboratory of Big Data Intelligent Computing China

Machine writing with large language models often relies on retrieval-augmented generation. However, these approaches remain confined within the boundaries of the model’s predefined scope, limiting the generation of content with rich information. Specifically, vanilla-retrieved information tends to lack depth, novelty, and suffers from redundancy, which negatively impacts the quality of generated articles, leading to shallow, unoriginal, and repetitive outputs. To address these issues, we propose OmniThink, a slow-thinking machine writing framework that emulates the human-like process of iterative expansion and reflection. The core idea behind OmniThink is to simulate the cognitive behavior of learners as they slowly deepen their knowledge of the topics. Experimental results demonstrate that OmniThink improves the knowledge density of generated articles without compromising metrics such as coherence and depth. Human evaluations and expert feedback further highlight the potential of OmniThink to address real-world challenges in the generation of long-form articles. Copyright © 2025, The Authors. 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

Functional Testing and Repair for ReRAM-Based Deep-Learning Accelerators

IEEE Design and Test

引用

IEEE Design and Test 2025年第3期42卷 66-73页

作者： Li, Wen Wang, Ying Li, Huawei Li, Xiaowei Zou, Kaiwei Shanxi University School of Computer and Information Technology Shanxi Taiyuan030006 China Institute of Big Data Science and Industry Shanxi University Shanxi Taiyuan030006 China State Key Laboratory of Processors Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China University of Chinese Academy of Sciences Beijing100049 China Tsinghua University Department of Electronic Engineering Beijing100084 China

Editor’s notes: This article offers an overview of the RRAMedy framework, designed to perform online testing, fault diagnosis, and fault mitigation in resistive random-access memory (ReRAM) devices, specifically in the context of ReRAM crossbar-based implementations for AI hardware acceleration. —Jyotika Athavale, Synopsys, USA —Haralampos-G. Stratigopoulos, Sorbonne Université, CNRS, LIP6, France. © 2013 IEEE.

关键词： Accuracy Fault tolerant systems Fault tolerance Deep learning Image edge detection Fault detection Computational modeling Training Reliability Hafnium

来源：评论

学校读者我要写书评

暂无评论

MSMAE-Net: multi-semantic and multi-attention enhanced network for image forgery localization

引用

Signal, Image and Video Processing 2025年第7期19卷 1-11页

作者： Liao, Jianjun Su, Lichao Lu, Menghan Engineering Research Center of Big Data Intelligence Ministry of Education Fuzhou China Fujian Key Laboratory of Network Computing and Intelligent Information Processing Fuzhou University Fuzhou China

The popularization of modern digital image technology has brought convenience to us, but it also poses many risks. The advancement of image editing software allows anyone to modify image content effortlessly. If these modified images are abused, they can severely impact societal safety and stability. To address these risks, we propose an end-to-end multi-scale collaborative enhancement image forgery localization network, termed MSMAE-Net. The method first employs a multi-branch feature extractor to initially capture both global and local information, with each branch optimized for different contextual information and features of the image. Next, to enhance the model’s ability to capture forged traces at various levels while maintaining the correlation between different spatial hierarchies, a boundary information aggregation module is designed. By constructing multiple branches with different receptive fields and collaborative learning with each other, a nested collaborative enhancement branch is designed to make the backbone network pay more attention to key features, so as to obtain better feature representation capability. Furthermore, to further fuse different semantic information, a novel strongly compatible semantic attention fusion module is proposed in this paper. Finally, an attention enhancement module is introduced in the localization of forgery regions to adaptively concentrate on key forged areas in the image and effectively capture the relationships between different pixels. In the experiments, we demonstrate that MSMAE-Net exhibits significant advantages in both localization accuracy and robustness in complex forgery scenarios when compared to other SOTA image forgery detection methods.

关键词：

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：