检索结果-南通市图书馆

A Simple yet Effective Framework for Active Learning to Rank

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Machine Intelligence Research 2024年第1期21卷 169-183页

作者： Qingzhong Wang haifang li Haoyi Xiong Wen Wang Jiang Bian Yu Lu Shuaiqiang Wang Zhicong Cheng Dejing Dou Dawei Yin Baidu Incorporated Beijing 100085China

While China has become the largest online market in the world with approximately 1 billion internet users,Baidu runs the world's largest Chinese search engine serving more than hundreds of millions of daily active users and responding to billions of queries per *** handle the diverse query requests from users at the web-scale,Baidu has made tremendous efforts in understanding users'queries,retrieving relevant content from a pool of trillions of webpages,and ranking the most relevant webpages on the top of the *** the components used in Baidu search,learning to rank(LTR)plays a critical role and we need to timely label an extremely large number of queries together with relevant webpages to train and update the online LTR *** reduce the costs and time con-sumption of query/webpage labelling,we study the problem of active learning to rank(active LTR)that selects unlabeled queries for an-notation and training in this ***,we first investigate the criterion-Ranking entropy(RE)characterizing the entropy of relevant webpages under a query produced by a sequence of online LTR models updated by different checkpoints,using a query-by-com-mittee(QBC)***,we explore a new criterion namely prediction variances(PV)that measures the variance of prediction res-ults for all relevant webpages under a *** empirical studies find that RE may favor low-frequency queries from the pool for la-belling while PV prioritizes high-frequency queries ***,we combine these two complementary criteria as the sample selection strategies for active *** experiments with comparisons to baseline algorithms show that the proposed approach could train LTR models to achieve higher discounted cumulative gain(i.e.,the relative improvement DCG4=1.38%)with the same budgeted labellingefforts.

关键词： Search information retrieval learning to rank active learning query by committee

低可见度环境下基于改进YOLOv3的井下人员定位方法

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

工矿自动化 2024年第9期50卷 130-137页

作者：路晓亚李海芳商丘工学院信息与电子工程学院河南商丘476000

煤矿井下光照不足、粉尘遮挡,井下视频监控系统采集的人员目标在二维图像中表现为小目标或低可见度目标时,原始YOLOv3网络的Darknet53特征金字塔结构无法充分提取和保留目标的细节信息,导致定位结果不准确。针对上述问题,提出了一种低... 详细信息

煤矿井下光照不足、粉尘遮挡,井下视频监控系统采集的人员目标在二维图像中表现为小目标或低可见度目标时,原始YOLOv3网络的Darknet53特征金字塔结构无法充分提取和保留目标的细节信息,导致定位结果不准确。针对上述问题,提出了一种低可见度环境下基于改进YOLOv3的井下人员定位方法。首先,结合β函数映射和帧间信息增强技术,提升低可见度环境下煤矿井下监控视频的清晰度。然后,采用更轻量级的Darknet-19替代YOLOv3中的Darknet53,并引入CIoU作为损失函数,利用改进YOLOv3识别增强后视频中的井下人员目标。最后,基于映射模型将识别到的目标从二维空间投影至三维空间,结合三维定位结果完成井下人员定位。选用某煤矿一段低可见度环境下井下监控视频进行实验,结果表明:①经过基于改进YOLOv3的井下人员定位方法处理后的视频帧亮度、可见度和各项评价指标(平均灰度、平均对比度、信息熵与灰度谱带宽)较原始视频均有明显提升,整体光照条件得到显著改善,且处理后的视频帧对比度得到增强,目标和背景之间更易区分,证明了采用的图像增强技术的有效性。②改进YOLOv3模型能准确识别视频帧中的井下工作人员,不存在漏识别问题。③采用已知位置的标定物或人工标注的的真实三维位置作为基准,计算投影结果与真实位置之间的偏差(偏差计算涵盖X,Y,Z方向上的距离偏差),其中X方向和Y方向上的偏差均小于0.2 m,Z方向上的偏差小于0.002 m,表明构建的映射模型的映射效果好且定位精度较高。

关键词：低可见度环境井下人员定位改进YOLOv3 三维空间定位 β函数映射帧间信息增强映射模型

Novel Model Using Kernel Function and Local Intensity Information for Noise Image Segmentation

同方期刊数据库博看期刊评论

在线全文

学校读者我要写书评

暂无评论

Tsinghua Science and Technology 2018年第3期23卷 303-314页

作者： Gang li haifang li ling Zhang College of Computer Science and Technology Taiyuan University of Technology

It remains a challenging task to segment images that are distorted by noise and intensity *** overcome these problems, in this paper, we present a novel region-based active contour model based on local intensity information and a kernel metric. By introducing intensity information about the local region, the proposed model can accurately segment images with intensity inhomogeneity. To enhance the model＇s robustness to noise and outliers, we introduce a kernel metric as its objective functional. To more accurately detect boundaries, we apply convex optimization to this new model, which uses a weighted total-variation norm given by an edge indicator function. Lastly, we use the split Bregman iteration method to obtain the numerical solution. We conducted an extensive series of experiments on both synthetic and real images to evaluate our proposed method, and the results demonstrate significant improvements in terms of efficiency and accuracy, compared with the performance of currently popular methods.

关键词： kernel metric image segmentation local intensity information convex optimization

link Prediction in Brain Networks Based on a Hierarchical Random Graph Model

在线全文

学校读者我要写书评

暂无评论

Tsinghua Science and Technology 2015年第3期20卷 306-315页

作者： Yanli Yang Hao Guo Tian Tian haifang li School of Computer Science and Technology Taiyuan University of Technology

link prediction attempts to estimate the likelihood of the existence of links between nodes based on available brain network information, such as node attributes and observed links. In response to the problem of the poor efficiency of general link prediction methods applied to brain networks, this paper proposes a hierarchical random graph model based on maximum likelihood estimation. This algorithm uses brain network data to create a hierarchical random graph model. Then, it samples the space of all possible dendrograms using a Markov-chain Monte Carlo algorithm. Finally, it calculates the average connection probability. It also employs an evaluation index. Comparing link prediction in a brain network with link prediction in three different networks （Treponemapallidum metabolic network, terrorist networks, and grassland species food webs） using the hierarchical random graph model, experimental results show that the algorithm applied to the brain network has the highest prediction accuracy in terms of AUC scores. With the increase of network scale, AUC scores of the brain network reach 0.8 before gradually leveling off. In addition, the results show AUC scores of various algorithms computed in networks of eight different scales in 28 normal people. They show that the HRG algorithm is far better than random prediction and the ACT global index, and slightly inferior to local indexes CN and LP. Although the HRG algorithm does not produce the best results, its forecast effect is obvious, and shows good time complexity.

关键词： brain network link prediction hierarchical random graph maximum likelihood estimation method

考虑平均应力影响的7075-T651铝合金疲劳性能研究

在线全文

学校读者我要写书评

暂无评论

机械强度 2024年第2期46卷 446-452页

作者：俞俊王宇李海芳翁硕何小田赵礼辉上海理工大学机械工程学院上海200093 北汽福田汽车股份有限公司时代领航卡车工厂潍坊262103 机械工业汽车机械零部件强度与可靠性评价重点实验室上海200093 上海市新能源汽车可靠性评价公共技术平台上海200093

研究了平均应力效应对7075-T651铝合金疲劳性能的影响,选取了R=0.5、0.3、0.1和-0.1四组应力比进行疲劳试验。结果表明,同一应力比下,随着峰值应力水平的增加,疲劳寿命逐渐降低;而同一峰值应力水平下,随着应力比的提高,疲劳寿命则呈现... 详细信息

研究了平均应力效应对7075-T651铝合金疲劳性能的影响,选取了R=0.5、0.3、0.1和-0.1四组应力比进行疲劳试验。结果表明,同一应力比下,随着峰值应力水平的增加,疲劳寿命逐渐降低;而同一峰值应力水平下,随着应力比的提高,疲劳寿命则呈现出逐渐升高的趋势。同时,使用扫描电子显微镜(Scanning Electron Microscopy,SEM)观察疲劳断口形貌发现,在峰值应力相同的情况下,裂纹萌生及扩展区的扇形面积随平均应力的增大而增大。最后,基于7075-T651铝合金疲劳实验结果,提出了一种考虑S-N-R的改进型预测模型,结果对比显示相较于其他模型,此模型在一定程度上对7075-T651铝合金具有更好的疲劳寿命预测效果。

关键词： 7075-T651铝合金平均应力疲劳寿命断口形貌预测模型

Forest soil CO_2 fluxes as a function of understory removal and N-fixing species addition

同方期刊数据库博看期刊评论

在线全文

学校读者我要写书评

暂无评论

Journal of Environmental Sciences 2011年第6期23卷 949-957页

作者： haifang li Shenglei Fu Hongting Zhao Hanping Xia Institute of Ecology South China Botanical Garden Chinese Academy of Sciences Guangzhou 510650 China College of Tourism Guilin University of Technology Guilin 541004 China E&E Applied Research LLC 3628 Oakwood Drive Longmont Colorado 80503 USA

We report on the effects of forest management practices of understory removal and N-fixing species（Cassia alata） addition on soil CO2 fluxes in an Eucalyptus urophylla plantation（EUp）,Acacia crassicarpa plantation（ACp）,10-species-mixed plantation（Tp）,and 30-species-mixed plantation（THp） using the static chamber method in southern *** forest management treatments,including（1） understory removal（UR）;（2） *** addition（CA）;（3） understory removal and replacement with ***（UR＋CA）;and（4） control without any disturbances（CK）,were applied in the above four forest plantations with three replications for each *** results showed that soil CO2 fluxes rates remained at a high level during the rainy season（from April to September）,followed by a rapid decrease after October reaching a minimum in *** CO2 fluxes were significantly higher（P 〈 0.01） in EUp（132.6 mg/（***）） and ACp（139.8 mg/（***）） than in Tp（94.0 mg/（***）） and THp（102.9 mg/（***））.Soil CO2 fluxes in UR and CA were significantly higher（P 〈 0.01） among the four treatments,with values of 105.7,120.4,133.6 and 112.2 mg/（***） for UR＋CA,UR,CA and CK,*** CO2 fluxes were positively correlated with soil temperature（P 〈 0.01）,soil moisture（P 〈 0.01）,NO3？-N（P 〈 0.05）,and litterfall（P 〈 0.01）,indicating that all these factors might be important controlling variables for soil CO2 *** study sheds some light on our understanding of soil CO2 flux dynamics in forest plantations under various management practices.

关键词： soil C02 fluxes forest management practices understory removal N-fixing species addition forest plantation

On-chip solid phase extraction coupled with electrophoresis using modified magnetic microspheres as stationary phase

在线全文

学校读者我要写书评

暂无评论

Science China Chemistry 2009年第12期52卷 2287-2294页

作者： li Hong li haifang CHEN ZhiFeng liN JinMing Department of Chemistry Tsinghua UniversityBeijing 100084China Chinese Academy of Inspection and Quarantine Beijing 100123China

A poly(dimethylsiloxane)(PDMS)/glass hybrid microchip for on-line solid phase extraction (SPE) and electrophoresis separation has been developed and evaluated. The SPE microchannel was crossed to the electrophoresis microchannel. All the microfluidic channels were etched on the glass substrate. The magnetic microspheres were coated with hydroxyl-terminated poly-dimethylsiloxane (PDMS-OH) serving as extraction phase,which could be conveniently immobilized into the sample pretreatment channel by magnetic field. The PDMS-OH microspheres were mobilized into and out of the pretreatment channel by injection flow. The 0.1 μmol/L solution of fluorescence isothiocyanate (FITC)-labeled phenylalanine (Phe) was electrically injected into the SPE channel and extracted onto the PDMS-OH microspheres bed. The enriched FITC-labeled Phe was electrically eluted by 9 mmol/L sodium acetate containing 10% acetonitrile and electrically driven into the electrophoresis channel and then separated. The preconcentration factor could reach 87.5 after sufficient extraction. A linear preconcentration curve was obtained with the initial FITC-labeled Phe concentration ranging from 6 nmol/L to 300 nmol/L (R2=0.9922) with 200 s loading time. The detection limit (S/N=3) for the FITC-labeled Phe was 3 nmol/L.

关键词： microfluidic device solid phase extraction magnetic microspheres hydroxyl-terminated poly-dimethylsiloxane

Combining Residual Attention Mechanisms and Generative Adversarial Networks for Hippocampus Segmentation

在线全文

学校读者我要写书评

暂无评论

Tsinghua Science and Technology 2022年第1期27卷 68-78页

作者： Hongxia Deng Yuefang Zhang Ran li Chunxiang Hu Zijian Feng haifang li the School of Information and Computer Taiyuan University of TechnologyTaiyuan 030600China

This research discussed a deep learning method based on an improved generative adversarial network to segment the *** convolutional configurations were proposed to capture information obtained by a segmentation *** addition,a generative adversarial network based on Pixel2Pixel was *** generator was a codec structure combining a residual network and an attention mechanism to capture detailed *** discriminator used a convolutional neural network to discriminate the segmentation results of the generated model and that of the *** the continuously transmitted losses of the generator and discriminator,the generator reached the optimal state of hippocampus segmentation.T1-weighted magnetic resonance imaging scans and related hippocampus labels of 130 healthy subjects from the Alzheimer’s disease Neuroimaging Initiative dataset were used as training and test data;similarity coefficient,sensitivity,and positive predictive value were used as evaluation *** showed that the network model could achieve an efficient automatic segmentation of the hippocampus and thus has practical relevance for the correct diagnosis of diseases,such as Alzheimer’s disease.

关键词： magnetic resonance imaging generative adversarial network residual network attention mechanism

A microchip to analyze single crystal growth and size-controllability

在线全文

学校读者我要写书评

暂无评论

Science China Chemistry 2009年第7期52卷 1014-1020页

作者： PANG YuanFeng liU JiangJiang li haifang liN JinMing The Key Laboratory of Bioorganic Phosphorus Chemistry and Chemical Biology Department of ChemistryTsinghua UniversityBeijing 100084China

A microfluidic device to control single crystallization on the micron scale has been developed. The salt solution was stored in the nano-volume gaps between the arrays of protrudent circular plots in the microchip. The mixed organic solvent was injected into the chip as the counter diffusion phase for crystallization forming. This device provides a liquid-liquid interface through which only one phase flows while the other stays at the fixed plot. Therefore, it is possible to control the position of crystallization on the fixed plot. We can control the size and the uniformity of single crystals from 5 to 50 μm in length by adjusting the relative factors, such as interface lifetime, breeds of the mix-organic solvents and injecting velocities. The longer interface lifetime and lower organic solvent injecting velocities can bring up larger and more asymmetric crystals, which nearly shows the same trend compared with the macroscopic crystallization. Finally, the effect of the surfactant on the crystallization in the microdevice was studied. By adding the surfactant into the liquid-liquid interface, smaller sizes of crystals can be obtained without changing the crystal configuration.

关键词： microfluidic device single crystallization liquid-liquid interface size controllability