检索结果-内蒙古大学图书馆

InFi: End-to-End Learning to Filter Input for resource-Efficiency in Mobile-Centric Inference

IEEE TrANSACTIONS ON MOBILE COMPUTING 2024年第5期23卷 3523-3538页

作者： Yuan, Mu Zhang, Lan He, Fengxiang Tong, Xueting Song, Miao-Hui Xu, Zhengyuan Li, Xiang-Yang Univ Sci & Technol China Sch Comp Sci & Technol Hefei 230052 Anhui Peoples R China Univ Sci & Technol China Hefei Comprehens Natl Sci Ctr Sch Comp Sci & Technol Sch Data SciInst Artificial Intelligence Hefei 230052 Anhui Peoples R China JD Com Inc JD Explore Acad Beijing 101111 Peoples R China Univ Sci & Technol China Inst Adv Technol Hefei 230052 Anhui Peoples R China Univ Sci & Technol China Key Lab Wireless Opt Commun Chinese Acad Sci Hefei 230052 Anhui Peoples R China

Mobile-centric AI applications have high requirements for the resource-efficiency of model inference. Input filtering is a promising approach to eliminate redundancy so as to reduce the cost of inference. Previous efforts have tailored effective solutions for many applications, but left two essential questions unanswered: (1) theoretical filterability of an inference workload to guide the application of input filtering techniques, thereby avoiding the trial-and-error cost for resource-constrained mobile applications;(2) robust discriminability of feature embedding to allow input filtering to be widely effective for diverse inference tasks and input content. To answer them, we first formulate the input filtering problem and theoretically compare the hypothesis complexity of inference models and input filters to understand the optimization potential. Then we propose the first end-to-end learnable input filtering framework that covers most state-of-the-art methods and surpasses them in feature embedding with robust discriminability. We design and implement InFi that supports different input modalities and mobile-centric deployments. Comprehensive evaluations confirm our theoretical results and show that InFi outperforms strong baselines in applicability, accuracy, and efficiency. InFi can achieve 8.5x throughput and save 95% bandwidth, while keeping over 90% accuracy, for a video analytics application on mobile platforms.

关键词： Input filtering model inference mobile computing multimodal data

来源：评论

学校读者我要写书评

暂无评论

data Imputation of Soil Pressure on Shield Tunnel Lining Based on random Forest Model

引用

SENSOrS 2024年第5期24卷 1560页

作者： Wang, Min Ye, Xiao-Wei Ying, Xin-Hong Jia, Jin-dian ding, Yang Zhang, di Sun, Feng Zhejiang Univ Polytech Inst Hangzhou 310058 Peoples R China Zhejiang Univ Dept Civil Engn Hangzhou 310058 Peoples R China Hangzhou City Univ Dept Civil Engn Hangzhou 310015 Peoples R China China Railway Siyuan Survey & Design Grp Co Ltd Wuhan 430063 Peoples R China

With the advancement of engineering techniques, underground shield tunneling projects have also started incorporating emerging technologies to monitor the forces and displacements during the construction and operation phases of shield tunnels. Monitoring devices installed on the tunnel segment components generate a large amount of data. However, due to various factors, data may be missing. Hence, the completion of the incomplete data is imperative to ensure the utmost safety of the engineering project. In this research, a missing data imputation technique utilizing random Forest (rF) is introduced. The optimal combination of the number of decision trees, maximum depth, and number of features in the rF is determined by minimizing the Mean Squared Error (MSE). Subsequently, complete soil pressure data are artificially manipulated to create incomplete datasets with missing rates of 20%, 40%, and 60%. A comparative analysis of the imputation results using three methods-median, mean, and rF-reveals that this proposed method has the smallest imputation error. As the missing rate increases, the mean squared error of the random Forest method and the other two methods also increases, with a maximum difference of about 70%. This indicates that the random forest method is suitable for imputing monitoring data.

关键词： shield tunnel soil pressure missing data imputation random forest

来源：评论

学校读者我要写书评

暂无评论

An advance in novel intelligent sensory technologies: From an implicit-tracking perspective of food perception

引用

COMPrEHENSIVE rEVIEWS IN FOOd SCIENCE ANd FOOd SAFETY 2024年第2期23卷 e13327-e13327页

作者： Zhao, Qian Ye, Zhiyue deng, Yong Chen, Jin Chen, Jianle Liu, donghong Ye, Xingqian Huan, Cheng Zhejiang Univ Fuli Inst Food Sci Coll Biosyst Engn & Food Sci Natl Local Joint Engn Res Ctr Intelligent Food Tec Hangzhou Peoples R China Zhejiang Univ Innovat Ctr Yangtze River Delta Jiaxing Peoples R China Zhejiang Univ Zhongyuan Inst Zhengzhou Peoples R China Zhejiang Univ Ningbo Innovat Ctr Ningbo Peoples R China Zhejiang Univ Fuli Inst Food Sci Coll Biosyst Engn & Food Sci Natl Local Joint Engn Lab Intelligent Food Technol Hangzhou 310058 Peoples R China

Food sensory evaluation mainly includes explicit and implicit measurement methods. Implicit measures of consumer perception are gaining significant attention in food sensory and consumer science as they provide effective, subconscious, objective analysis. A wide range of advanced technologies are now available for analyzing physiological and psychological responses, including facial analysis technology, neuroimaging technology, autonomic nervous system technology, and behavioral pattern measurement. However, researchers in the food field often lack systematic knowledge of these multidisciplinary technologies and struggle with interpreting their results. In order to bridge this gap, this review systematically describes the principles and highlights the applications in food sensory and consumer science of facial analysis technologies such as eye tracking, facial electromyography, and automatic facial expression analysis, as well as neuroimaging technologies like electroencephalography, magnetoencephalography, functional magnetic resonance imaging, and functional near-infrared spectroscopy. Furthermore, we critically compare and discuss these advanced implicit techniques in the context of food sensory research and then accordingly propose prospects. Ultimately, we conclude that implicit measures should be complemented by traditional explicit measures to capture responses beyond preference. Facial analysis technologies offer a more objective reflection of sensory perception and attitudes toward food, whereas neuroimaging techniques provide valuable insight into the implicit physiological responses during food consumption. To enhance the interpretability and generalizability of implicit measurement results, further sensory studies are needed. Looking ahead, the combination of different methodological techniques in real-life situations holds promise for consumer sensory science in the field of food research.

关键词： consumer science facial analysis food flavor perception intelligent sensory technology neuroimaging technology

来源：评论

学校读者我要写书评

暂无评论

Preparation of a GO@PVP@Ag SErS Substrate and Its Application in the detection of Phenanthylene in Aqueous Solution

引用

COATINGS 2025年第3期15卷

作者： Liu, Junyu Li, Hansheng Cui, Wencan Sang, Shihua Guan, Liang Zhao, Chen Chengdu Univ Technol Coll Mat & Chem & Chem Engn Chengdu 610059 Peoples R China Army Logist Acad PLA Dept Basic Courses Chongqing 401331 Peoples R China

This study focuses on the preparation of a highly active surface-enhanced raman spectroscopy (SErS) substrate, where Ag nanoparticles synthesized by the Lee-Meisel method are loaded onto the surface of GO@PVP nanoparticles through classical electrostatic attraction. The substrate was characterized by SEM and XPS analysis. The SErS-active substrate was utilized for the detection of phenanthrene in aqueous solutions, achieving a detection limit of 10-8 g/L (5.6 x 10-11 mol/L). In comparison to conventional raman spectroscopy, this method exhibits a significant reduction in the detection limit and holds promise for trace polycyclic aromatic hydrocarbons (PAHs) monitoring in environmental samples.

关键词： polycyclic aromatic hydrocarbons SErS GO@PVP@Ag substrate graphene oxide

来源：评论

学校读者我要写书评

暂无评论

remnants and fragments of the subducted paleo-Pacific plate:Constraints from geochemistry and geophysics

引用

Science china Earth Sciences 2024年第10期67卷 3041-3061页

作者： Xisheng XU Zhouchuan HUANG dingsheng JIANG Gang ZENG Li-Qun dAI State Key Laboratory for Mineral Deposits Research School of Earth Sciences and EngineeringNanjing UniversityNanjing 210023China School of Earth and Space Sciences University of Science and Technology of ChinaHefei 230026China

The subduction and rollback of the paleo-Pacific plate during Mesozoic time was the key engine for the evolution of the continental margin in eastern china. It led to lateral accretion of continental crust in Northeast china, lithospheric destruction beneath the North china Craton, and the generation of huge volumes of felsic magmatic rocks in South china. This had a profound influence on deep material cycles and the evolution of epigenetic environmental systems along the continental margin of East Asia. To fully understand the transformation of the dynamic mechanism during the subduction and rollback of the paleoPacific plate, we have attempted to trace the remnants and fragments of the subducted paleo-Pacific plate at great depths. Such remnants in both temporal and spatial dimensions can be tracked by using geochemical and geophysical approaches. Studies of the trace elements, Mg-Zn isotopes and Os-Nd-Hf-Pb-O isotopes in continental basalts from eastern china reveal a significant number of the remnants of subduction of the paleo-Pacific plate, and the initial subduction can be traced back to the Early Jurassic. Large-scale geophysical imaging unveils a multitude of high-velocity anomalies in the lower mantle of East ***, many high-velocity bodies, aptly referred to as “slab graveyards”, are nestled at the base of the lower mantle. Numerous isolated high-velocity anomalies are also present in the upper part of the lower mantle, creating conduits for the descent of the subducted slabs into the lower mantle. However, a resolution of the remnants for the subducted slabs within the lower mantle are quite low. Consequently, their impact on the lower mantle's dynamics is yet to be thoroughly investigated. Finally, the presently observed big mantle wedge(BMW) in East Asia has developed through subduction of the Pacific plate in the ***, following the rollback of the paleo-Pacific plate(began at ~145 Ma), a Cretaceous BMW system would also form above

关键词： Paleo-Pacific plate Big mantle wedge Late Mesozoic remnants and fragments Geochemistry Geophysics

来源：评论

学校读者我要写书评

暂无评论

Corrosion Behavior of Nacre-Inspired (TiBw-TiB2)/Al Composites Fabricated by Freeze Casting

引用

MATErIALS 2024年第11期17卷 2534页

作者： Zhang, Jidong Qian, Mingfang Yang, ruiqing Yu, Feng Zhang, Xuexi Jia, Zhenggang Li, Aibin Wang, Guisong Geng, Lin Harbin Inst Technol Sch Mat Sci & Engn Harbin 150001 Peoples R China Harbin Inst Technol Natl Key Lab Precis Hot Forming Met Harbin 150001 Peoples R China Northeast Light Alloy Co Ltd Harbin 150060 Peoples R China Harbin Inst Technol State Key Lab Adv Welding & Joining Harbin 150001 Peoples R China

Nacre-inspired metal matrix composites have received much attention due to their excellent deformation coordination ability, which can achieve the synergy of strength and ductility. The preparation of nacre-like Al matrix composites by freeze casting has been a promising application, but the continuous ceramic-rich layer affects the corrosion resistance of the composites, facing complex corrosion problems during service. In this work, the microstructure and corrosion behavior of the nacre-inspired (TiBw-TiB2)/Al composites fabricated by freeze casting and squeeze casting were systematically studied. The results indicated that the Al layers and ceramic-rich layers had little change, about 35 mu m and 31 mu m, respectively, with an increasing ratio of the Ti/TiB2. Meanwhile, a high Ti/TiB2 ratio resulted in an increase in the Fe-Ti intermetallic phases, which was detrimental to the corrosion performance of the composites and was prone to pitting. The electrochemical test results showed that the 3Ti7TiB2 composite had the lowest corrosion current density (15.9 mu A) and intergranular corrosion depth (231 mu m), indicating that it had the best corrosion resistance, which can be attributable to its stable and dense passivation film. Two different corrosion phenomena during the intergranular corrosion test existed in the present nacre-inspired (TiBw-TiB2)/Al composites: intergranular corrosion in the Al matrix layer and pitting corrosion in the ceramic-rich layer. Among all the composites, the corrosion depth of the 3Ti7TiB(2 )composite was the smallest and significantly less than that of the 2024Al alloy. In addition, the continuous ceramic-rich layer acted as a corrosion channel during corrosion, significantly degrading the corrosion resistance of the nacre-like Al composites.

关键词： freeze casting nacre-like architecture (TiBw-TiB2)/Al composites corrosion behavior

来源：评论

学校读者我要写书评

暂无评论

research on High-Stability Composite Control Methods for Telescope Pointing Systems under Multiple disturbances

引用

SENSOrS 2024年第9期24卷 2907页

作者： Zhang, rui Zhao, Kai Fang, Sijun Fan, Wentong Hai, Hongwen Luo, Jian Li, Bohong Sun, Qicheng Song, Jie Yan, Yong Sun Yat sen Univ MOE Key Lab TianQin Mission TianQin Res Ctr Gravitat Phys Zhuhai Campus Zhuhai 519082 Peoples R China Sun Yat sen Univ Frontiers Sci Ctr TianQin Gravitat Wave Res Ctr CNSA Sch Phys & Astron Zhuhai Campus Zhuhai 519082 Peoples R China

during the operation of space gravitational wave detectors, the constellation configuration formed by three satellites gradually deviates from the ideal 60 degrees angle due to the periodic variations in orbits. To ensure the stability of inter-satellite laser links, active compensation of the breathing angle variation within the constellation plane is achieved by rotating the optical subassembly through the telescope pointing mechanism. This paper proposes a high-performance robust composite control method designed to enhance the robust stability, disturbance rejection, and tracking performance of the telescope pointing system. Specifically, based on the dynamic model of the telescope pointing mechanism and the disturbance noise model, an H-infinity controller has been designed to ensure system stability and disturbance rejection capabilities. Meanwhile, employing the method of an H-infinity norm optimized disturbance observer (HOdOB) enhances the nonlinear friction rejection ability of the telescope pointing system. The simulation results indicate that, compared to the traditional disturbance observer (dOB) design, utilizing the HOdOB method can enhance the tracking accuracy and pointing stability of the telescope pointing system by an order of magnitude. Furthermore, the proposed composite control method improves the overall system performance, ensuring that the stability of the telescope pointing system meets the 10 nrad/Hz(1/2) @0.1 mHz similar to 1 Hz requirement specified for the TianQin mission.

关键词： space gravitational wave breathing angle variation telescope pointing mechanism H-infinity controller HOdOB pointing stability

来源：评论

学校读者我要写书评

暂无评论

Multimodal Physiological Analysis of Impact of Emotion on Cognitive Control in Vr

引用

IEEE TrANSACTIONS ON VISUALIZATION ANd COMPUTEr GrAPHICS 2024年第5期30卷 2044-2054页

作者： Li, Ming Pan, Junjun Li, Yu Gao, Yang Qin, Hong Shen, Yang Beihang Univ Beijing Peoples R China Beijing Normal Univ Beijing Peoples R China SUNY Stony Brook Stony Brook NY USA

Cognitive control is often perplexing to elucidate and can be easily influenced by emotions. Understanding the individual cognitive control level is crucial for enhancing Vr interaction and designing adaptive and self-correcting Vr/Ar applications. Emotions can reallocate processing resources and influence cognitive control performance. However, current research has primarily emphasized the impact of emotional valence on cognitive control tasks, neglecting emotional arousal. In this study, we comprehensively investigate the influence of emotions on cognitive control based on the arousal-valence model. A total of 26 participants are recruited, inducing emotions through Vr videos with high ecological validity and then performing related cognitive control tasks. Leveraging physiological data including EEG, HrV, and EdA, we employ classification techniques such as SVM, KNN, and deep learning to categorize cognitive control levels. The experiment results demonstrate that high-arousal emotions significantly enhance users' cognitive control abilities. Utilizing complementary information among multi-modal physiological signal features, we achieve an accuracy of 84.52% in distinguishing between high and low cognitive control. Additionally, time-frequency analysis results confirm the existence of neural patterns related to cognitive control, contributing to a better understanding of the neural mechanisms underlying cognitive control in Vr. Our research indicates that physiological signals measured from both the central and autonomic nervous systems can be employed for cognitive control classification, paving the way for novel approaches to improve Vr/Ar interactions.

关键词： Cognitive control emotion physiological signal deep learning

来源：评论

学校读者我要写书评

暂无评论

Embedding-Alignment Fusion-Based Graph Convolution Network With Mixed Learning Strategy for 4d Medical Image reconstruction

引用

IEEE JOUrNAL OF BIOMEdICAL ANd HEALTH INFOrMATICS 2024年第5期28卷 2916-2929页

作者： Li, Jingshu Fu, Tianyu Song, Hong Fan, Jingfan Xiao, deqiang Lin, Yucong Gu, Ying Yang, Jian Beijing Inst Technol Sch Opt & Photon Beijing 100081 Peoples R China Beijing Inst Technol Sch Med Technol Beijing 100081 Peoples R China Beijing Inst Technol Sch Comp Sci & Technol Beijing 100081 Peoples R China Chinese Peoples Liberat Army Gen Hosp Med Ctr 1 Dept Laser Med Beijing 100853 Peoples R China Chinese Acad Med Sci Precis Laser Med Diag & Treatment Innovat Unit Beijing 100000 Peoples R China

In recent years, 4d medical image involving structural and motion information of tissue has attracted increasing attention. The key to the 4d image reconstruction is to stack the 2d slices based on matching the aligned motion states. In this study, the distribution of the 2d slices with the different motion states is modeled as a manifold graph, and the reconstruction is turned to be the graph alignment. An embedding-alignment fusion-based graph convolution network (GCN) with a mixed-learning strategy is proposed to align the graphs. Herein, the embedding and alignment processes of graphs interact with each other to realize a precise alignment with retaining the manifold distribution. The mixed strategy of self- and semi-supervised learning makes the alignment sparse to avoid the mismatching caused by outliers in the graph. In the experiment, the proposed 4d reconstruction approach is validated on the different modalities including Computed Tomography (CT), Magnetic resonance Imaging (MrI), and Ultrasound (US). We evaluate the reconstruction accuracy and compare it with those of state-of-the-art methods. The experiment results demonstrate that our approach can reconstruct a more accurate 4d image.

关键词： 4d image reconstruction graph embedding manifold alignment slice stacking

来源：评论

学校读者我要写书评

暂无评论

Hyperspectral Image Classification Using Spectral-Spatial double-Branch Attention Mechanism

引用

rEMOTE SENSING 2024年第1期16卷 193页

作者： Kang, Jianfang Zhang, Yaonan Liu, Xinchao Cheng, Zhongxin Natl Cryosphere Desert Data Ctr Lanzhou 730000 Peoples R China Chinese Acad Sci Northwest Inst Ecoenvironm & Resources Lanzhou 730000 Peoples R China Lanzhou Univ Sch Informat Sci & Engn Lanzhou 730000 Peoples R China

In recent years, deep learning methods utilizing convolutional neural networks have been extensively employed in hyperspectral image classification (HSI) applications. Nevertheless, while a substantial number of stacked 3d convolutions can indeed achieve high classification accuracy, they also introduce a significant number of parameters to the model, resulting in inefficiency. Furthermore, such intricate models often exhibit limited classification accuracy when confronted with restricted sample data, i.e., small sample problems. Therefore, we propose a spectral-spatial double-branch network (SSdBN) with an attention mechanism for HSI classification. The SSdBN is designed with two independent branches to extract spectral and spatial features, respectively, incorporating multi-scale 2d convolution modules, long short-term memory (LSTM), and an attention mechanism. The flexible use of 2d convolution, instead of 3d convolution, significantly reduces the model's parameter count, while the effective spectral-spatial double-branch feature extraction method allows SSdBN to perform exceptionally well in handling small sample problems. When tested on 5%, 0.5%, and 5% of the Indian Pines, Pavia University, and Kennedy Space Center datasets, SSdBN achieved classification accuracies of 97.56%, 96.85%, and 98.68%, respectively. Additionally, we conducted a comparison of training and testing times, with results demonstrating the remarkable efficiency of SSdBN.

关键词： convolutional neural network hyperspectral image classification multi-scale 2d convolution LSTM attention mechanism

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：