检索结果-内蒙古大学图书馆

8th International Conference on Instrumentation and Measurement, Computer, Communication and Control (IMCCC)

作者： Li, Yunbo Ding, Yu Bai, Wei Jiao, Shanshan Pan, Zhisong Army Engn Univ PLA Coll Command & Informat Syst Nanjing 211107 Jiangsu Peoples R China

ISBN: (纸本)9781538682463

In this paper, we propose a scalable network for tiny object detection based on Faster RCNN. Compared with the previous feature extraction network, our network can be better applied to tiny objects. In the process of feature extraction, the feature representation of large object will be strengthened, and the important tiny object information is ignored. By merging the feature maps output from different filters on the same layer, different sizes of targets will be captured. Then, not only considers the width of the network, but also realizes the deep integration of the network, which can avoid that the network is too deep to filter out tiny target information. Finally, by optimizing the algorithm for tiny objects based on deep learning, we achieved the best results with the accuracy rate of 34.1% on the Tsinghua-Tencent 100K.

关键词： tiny object detection Faster Rcnn feature extraction algorithm optimization deep learning

来源：评论

学校读者我要写书评

暂无评论

Genetic optimization of Training Sets for Improved Machine Learning Models of Molecular Properties

引用

JOURNAL OF PHYSICAL CHEMISTRY LETTERS 2017年第7期8卷 1351-1359页

作者： Browning, Nicholas J. Ramakrishnan, Rapunathan von Lilienfeld, O. Anatole Roethlisberger, Ursula Ecole Polytech Fed Lausanne Swiss Fed Inst Technol Inst Chem Sci & Engn Lab Computat Chem & Biochem CH-1015 Lausanne Switzerland Univ Basel Natl Ctr Computat Design & Discovery Novel Mat MA Klingelbergstr 80 CH-4056 Basel Switzerland Univ Basel Dept Chem Inst Phys Chem Klingelbergstr 80 CH-4056 Basel Switzerland Tata Inst Fundamental Res Ctr Interdisciplinary Sci 21 Brundavan Colony Hyderabad 500075 Andhra Prades India

The training of molecular models of quantum mechanical properties based on statistical machine learning requires large data sets which exemplify the map from chemical structure to molecular property. Intelligent a priori selection of training examples is often difficult or impossible to achieve, as prior knowledge may be unavailable. Ordinarily representative selection of training molecules from such data sets is achieved through random sampling. We use genetic algorithms for the optimization of training set composition consisting of tens of thousands of small organic molecules. The resulting machine learning models are considerably more accurate: in the limit of small training sets, mean absolute errors for out-of-sample predictions are reduced by up to similar to 75%. We discuss and present optimized training sets consisting of 10 molecular classes for all molecular properties studied. We show that these classes can be used to design improved training sets for the generation of machine learning models of the same properties in similar but unrelated molecular sets.

关键词： algorithm optimization molecular model SETTING mean absolute error Machine Learning Dataset molecular properties Random sampling

来源：评论

学校读者我要写书评

暂无评论

Research Progress of Intelligent optimization algorithm in Big Data Background

Research Progress of Intelligent Optimization Algorithm in B...

引用

2nd International Conference on Artificial Intelligence and Engineering Applications(AIEA 2017)

作者： DUO ZHANG College of Management and Economics Tianjin University

Big data is the inevitable outcome of the rapid development of modern information technology, effective analysis and processing of big data will not only bring great economic value, but will also promote social development. Under the big data environment, the data scale, speed of emergence and its difficulty make optimizing issues very complex. In recent decades, genetic algorithm(GA), particle swarm optimization(PSO), ant colony optimization(ACO), Artificial Fish School algorithm, Bacteria Foraging optimization algorithm(BFOA), artificial neural networks(ANNs) and other multi-population intelligent algorithms appeared. In this paper, several typical intelligent optimization algorithms are introduced, including genetic algorithm, particle swarm optimization algorithm, ant colony algorithm, artificial fish swarm algorithm and bacterial foraging algorithm. The basic principles of five algorithms are described respectively, along with the direction of improvement and feasible applications.

关键词： Big data intelligent optimization algorithm algorithm optimization genetic algorithm

来源：评论

学校读者我要写书评

暂无评论

Genetic algorithm optimization for Codewords Correction in MIMO Broadcast Channels

Genetic Algorithm Optimization for Codewords Correction in M...

引用

IEEE International Midwest Symposium on Circuits and Systems

作者： Mouncef Benmimoune Daniel Massicotte Sebastien Roy Electrical and Computer Engineering Department Universite du Quebec a Trois-Rivieres Laval University

ISBN: (纸本)9781467325264

This paper presents a novel precoding approach for MIMO broadcast channels, in which is performed on both sides of the wireless link. The aim of the proposed approach is to avoid the same precoding vectors choice when all users use a common codebook. In the proposal approach, firstly, we focus on the brute codeword selection at the transmitter side, thus there is zero probability to choose the same vector for more than one user. However, this solution leads an exhaustive search, especially when the number of user and the codebook size increase. To overcome this issue, secondly, we adopt the genetic algorithm in order to reduce the codeword search complexity. Compared with zero-forcing beamforming (ZFBF), the conducted simulation results for the critical scenario of low SNR, show that our scheme is better than ZFBF with the assumption of both perfect and partial channel state information at the transmitter.

关键词： broadcast channels algorithm optimization Genetic algorithm (GA) Channel state information multiple input multiple output systems Transmitters codeword codebook low SNR prediction coding exhaustive search Consumers Transmitter

来源：评论

学校读者我要写书评

暂无评论

Estimation of Leaf Nitrogen Content in Wheat Using New Hyperspectral Indices and a Random Forest Regression algorithm

引用

REMOTE SENSING 2018年第12期10卷 1940-1940页

作者： Liang, Liang Di, Liping Huang, Ting Wang, Jiahui Lin, Li Wang, Lijuan Yang, Minhua Jiangsu Normal Univ Sch Geog Geomat & Planning Xuzhou 221116 Jiangsu Peoples R China George Mason Univ Ctr Spatial Informat Sci & Syst Fairfax VA 22030 USA Cent South Univ Sch Geosci & Infophys Changsha 410083 Hunan Peoples R China

Novel hyperspectral indices, which are the first derivative normalized difference nitrogen index (FD-NDNI) and the first derivative ratio nitrogen vegetation index (FD-SRNI), were developed to estimate the leaf nitrogen content (LNC) of wheat. The field stress experiments were conducted with different nitrogen and water application rates across the growing season of wheat and 190 measurements were collected on canopy spectra and LNC under various treatments. The inversion models were constructed based on the dataset to evaluate the ability of various spectral indices to estimate LNC. A comparative analysis showed that the model accuracies of FD-NDNI and FD-SRNI were higher than those of other commonly used hyperspectral indices including mNDVI(705), mSR, and NDVI705, which was indicated by higher R-2 and lower root mean square error (RMSE) values. The least squares support vector regression (LS-SVR) and random forest regression (RFR) algorithms were then used to optimize the models constructed by FD-NDNI and FD-SRNI. The p-R-2 values of the FD-NDNI_RFR and FD-SRNI_RFR models reached 0.874 and 0.872, respectively, which were higher than those of the exponential and SVR model and indicated that the RFR model was accurate. Using the RFR inversion model, remote sensing mapping for the Operative Modular Imaging Spectrometer (OMIS) image was accomplished. The remote sensing mapping of the OMIS image yielded an accuracy of R-2 = 0.721 and RMSE = 0.540 for FD-NDNI and R-2 = 0.720 and RMSE = 0.495 for FD-SRNI, which indicates that the similarity between the inversion value and the measured value was high. The results show that the new hyperspectral indices, i.e., FD-NDNI and FD-SRNI, are the optimal hyperspectral indices for estimating LNC and that the RFR algorithm is the preferred modeling method.

关键词： hyperspectral remote sensing crop parameter inversion spectral index design derivative algorithm optimization

来源：评论

学校读者我要写书评

暂无评论

FEASIBILITY OF IRIS IDENTIFICATION algorithm optimization BY FRACTIONAL TEMPLATE MATCHING

FEASIBILITY OF IRIS IDENTIFICATION ALGORITHM OPTIMIZATION BY...

引用

IET Conference on Image Processing

作者： S. D. Winchell R. N. Rakvic Electrical and Computer Engineering Department United States Naval Academy

ISBN: (纸本)9781622762675

The human iris exhibits random and unique textural patterns that allow for identification with high accuracy. These patterns are evident in near-infrared (NIR) imagery, even for very dark irises. Iris templates are created from NIR iris imagery with the Ridge Energy Direction (RED) recognition algorithm and subsequently matched to measure performance. In this paper we investigate the feasibility of algorithm optimization by performing an initial fractional template comparison to eliminate high Hamming distance (HD) score matches followed by a full template re-comparison or iterative higher fractional template comparison on remaining templates. Recognition performance for different fractional template areas is analyzed with a view towards substantial improvement of algorithm identification time performance. The feasibility of identification time reduction by 60% or more is reported both ICE and Bath data sets.

关键词： Iris recognition Ridge Energy Direction (RED) Hamming Distance algorithm Performance Template Matching jigs(templates) algorithm optimization Ice algorithms INTEGRATED RISK INFORMATION SYSTEM Random Imagery (Psychotherapy) images

来源：评论

学校读者我要写书评

暂无评论

The optimization and Implementation of the Auto-Exposure algorithm Based on Image Entropy 8

The Optimization and Implementation of the Auto-Exposure Alg...

引用

8th International Conference on Digital Image Processing (ICDIP)

作者： Ning, Jingyi Lu, Tiejun Liu, Liyan Guo, Liye Jin, Xiaofeng 2 Siyingmen N Rd Beijing 100076 Peoples R China

ISBN: (纸本)9781510605039;9781510605046

To achieve auto-exposure in digital cameras, image brightness is widely used because of its direct relationship with exposure time. Although image entropy which represents information measure of image is widely used in various image processing applications, it has not been used on FPGA in AE system due to its complex calculation. Image with maximum entropy value contains more information. So this paper presents an algorithm based on image entropy. By searching maximum entropy value, the image can get an appreciate exposure. What's more, by using formula manipulation of image entropy and piecewise linearization of the log function, the optimized algorithm grasps the overall change rule in stead of traditional calculation and has been realized on FPGA. The experiment results show that on basis of 10M work frequency of CMOS image sensor ( electronic shutter, 1024 by 1024 pixels, on-chip 12-bit ADC) and 100M clock frequency of FPGA (ALTERA-EP2S60F48414N), this algorithm works well. And at the same time, the algorithm improves the amount of image information and increases the accuracy of auto exposure.

关键词： auto-exposure image-entropy algorithm optimization realize on FPGA

来源：评论

学校读者我要写书评

暂无评论

Real-time phase demodulation and data administration of distributed optical fiber vibration sensing system 14

Real-time phase demodulation and data administration of dist...

引用

Conference on Fiber Optic Sensors and Applications XIV

作者： Qin, Mengzhe He, Xiangge Liu, Fei Zheng, Xiaoping Zhang, Min Tsinghua Univ Dept Elect Engn Beijing 100084 Peoples R China Peking Univ Inst Ocean Res Beijing 100871 Peoples R China

ISBN: (数字)9781510609181

ISBN: (纸本)9781510609181;9781510609174

Distributed optical fiber sensing system (DOFS) has great potential in areas of petroleum exploration and ocean defense. By algorithm optimization in different coding environment, the multi-point and real-time heterodyne demodulation of DOFS is achieved. In experiments, the length of the optical fiber is 500m, the spatial resolution is 5m and the system sampling rate is 200kHz, under which condition the data rate reaches up to 160MB/s and the system can stilled be demodulated timely. Based on this, by plotting the three-dimensional diagram (vibration intensity versus time and space), the whole DOFS can be detected continuously and accurately.

关键词： DOFS heterodyne demodulation algorithm optimization three-dimensional diagram

来源：评论

学校读者我要写书评

暂无评论

Fast and Low-Power Behavior Analysis on Vehicles Using Smartphones 6

Fast and Low-Power Behavior Analysis on Vehicles Using Smart...

引用

6th IEEE International Symposium on Next-Generation Electronics (ISNE)

作者： Xu, Xiaoqing Yin, Shouyi Ouyang, Peng Tsinghua Univ Inst Microelect Beijing Peoples R China

ISBN: (纸本)9781538630969

More and more deep learning methods are applied in unmanned or assisted driving, and have achieved very excellent performance. This paper describes long short-term memory recurrent neural networks used in assisted driving, which can capture the long temporal dependencies of multiple vehicles sensors' data, supporting drivers' behavior analysis on vehicles. Some optimization methods, such as model compression, weight quantization, adaptive window segmentation, are applied to make the deep network faster and less power. Therefore, it can be easily deployed on smart-phones and other embedded devices due to its moderate energy consumption and low latency. The architecture was trained in a sequence-to-sequence prediction manner, and it explicitly learns to predict the driving patterns given the temporal context. The experiment is executed on the smart-phone. Experimental results for different parameters are also presented in the paper. At last, we reduce the model size to 77 KB, the processing time to 4.27 ms, and the power overhead is 7.7 mW, the percentage of improved performance by our optimizations is over 60%.

关键词： deep learning drivers' behavior analysis algorithm optimization

来源：评论

学校读者我要写书评

暂无评论

Research on Scene Management Technology Based on Adaptive Binary Tree and Scene Graph

Research on Scene Management Technology Based on Adaptive Bi...

引用

IEEE International Conference on Information and Automation (ICIA)

作者： Su, Mo Guo, Ruifeng Wang, Hongliang Wang, Shuai Univ Chinese Acad Sci Beijing Peoples R China Chinese Acad Sci Shenyang Inst Comp Technol Shenyang Liaoning Peoples R China Shandong Acad Sci Informat Res Inst Jinan Shandong Peoples R China

ISBN: (纸本)9781538631546

Scene management technology is one of the key technologies of virtual reality and visualization. In this paper, we propose a new method based on adaptive binary tree (ABT) and scene graph, which is used to improve the real-time rendering of indoor and outdoor objects and enhance the organization efficiency of scenes structure. The generation algorithm of adaptive binary tree, the scoring standard of the splitting plane, the algorithm of search effective segmentation of plane and related algorithms are described in detail. Due to the characteristics of high accuracy of adaptive binary tree space subdivision and strong adaptability of scene graph, the paper proposes the new management model that combines adaptive binary tree space subdivision algorithm to scene graph, forming the strategy of scene management. It not only expands the range of application of scene organization, but also can help to improve the subsequent rendering efficiency. The experimental results have demonstrated that our method greatly shortens the three-dimensional scene organization time, and accelerates real-time rendering speed of complex scene.

关键词： adaptive binary tree scene graph scene management spatial subdivision algorithm optimization

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：