检索结果-内蒙古大学图书馆

13th IEEE International Symposium on Parallel and Distributed Processing with Applications

作者： Karlsson, Andreas Sohl, Joar Liu, Dake Linkoping Univ Dept Elect Engn S-58183 Linkoping Sweden

ISBN: (纸本)9781467379526

This paper presents the novel heterogeneous DSP architecture ePUMA and demonstrates its features through an implementation of sorting of larger data sets. We derive a sorting algorithm with fixed-size merging tasks suitable for distributed memory architectures, which allows very simple scheduling and predictable data-independent sorting time. The implementation on ePUMA utilizes the architecture's specialized compute cores and control cores, and local memory parallelism, to separate and overlap sorting with data access and control for close to stall-free sorting. Penalty-free unaligned and out-of-order local memory access is used in combination with proposed application-specific sorting instructions to derive highly efficient local sorting and merging kernels used by the system-level algorithm. Our evaluation shows that the proposed implementation can rival the sorting performance of high-performance commercial CPUs and GPUs, with two orders of magnitude higher energy efficiency, which would allow high-performance sorting on low-power devices.

关键词： DSP ePUMA sorting sorting programs grade Digital signal processing Desmoplakin Gene distributed memory architecture Local Memory energy component Energy efficiency sorting algorithm ARCHITECTURE Access to Information information retrieval

来源：评论

学校读者我要写书评

暂无评论

The effect of class distribution on classification algorithms in credit risk assessment

The effect of class distribution on classification algorithm...

引用

International Convention on Information and Communication Technology, Electronics and Microelectronics

作者： K. Andric D. Kalpic Privredna Banka Zagreb Radni?ka 50 10000 Zagreb Hrvatska Sveu?ili?te u Zagrebu Fakultet elektrotehnike i ra?unarstva Unska 3 10000 Zagreb Hrvatska

ISBN: (纸本)9781509025435

This paper makes a practical contribution to classification in credit risk assessment by providing empirical evidence on which class distribution in the training sample should be used to maximize performance of different classification algorithms. Data used for credit risk assessment are often imbalanced due to the number of defaulting clients, who represent the minority class, being much less numerous than the non-defaulting majority class. Classification algorithms are usually biased towards the majority class and can show deceivingly high overall prediction accuracy, while at the same time exhibiting poor performance in prediction of the minority class. Although altering class distribution can be an effective method for alleviating the adverse impact of class imbalance, limited research efforts have been devoted to empirical evaluation of the role of class distribution in credit risk assessment, especially when dealing with real life data samples. To address this issue, an empirical study on the effect of different proportion of training examples belonging to each class on classifiers' performance is presented. In addition to logistic regression, neural networks and gradient boosting methods were evaluated using several real life and publicly available datasets from the UCI Machine Learning Repository.

关键词： Classification algorithms Risk management Logistics Boosting Neurons Training Biological neural networks sorting algorithm Risk Management Biological neural networks Logistics Boost Risk Assessment Functional training Neurons Power grids

来源：评论

学校读者我要写书评

暂无评论

Induced sorting Suffixes in External Memory with Better Design and Less Space 1

引用

22nd International Symposium on String Processing and Information Retrieval (SPIRE)

作者： Liu, Wei Jun Nong, Ge Chan, Wai Hong Wu, Yi Sun Yat Sen Univ Guangzhou 510275 Guangdong Peoples R China Gannan Normal Univ Ganzhou Peoples R China SYSU CMU Shunde Int Joint Res Inst Guangzhou Guangdong Peoples R China Hong Kong Inst Educ Tai Po Hong Kong Peoples R China

ISBN: (数字)9783319238265

ISBN: (纸本)9783319238265;9783319238258

Recently, several attempts have been made to extend the internal memory suffix array (SA) construction algorithm SA-IS to the external memory model, e.g., eSAIS, EM-SA-DS and DSA-IS. While the developed programs for these algorithms achieve remarkable performance in terms of I/O complexity and speed, their designs are quite complex and their disk requirements remain rather heavy. Currently, the core algorithmic part of each of these programs consists of thousands of lines in C++, and the average peak disk requirement is over 20n bytes for an input string of size n < 2(40). We re-investigate the problem of induced sorting suffixes in external memory and propose a new algorithm SAIS-PQ (SAIS with Priority Queue) and its enhanced alternative SAIS-PQ+. Using the library STXXL, the core algorithmic parts of SAIS-PQ and SAIS-PQ+ are coded in around 800 and 1600 lines in C++, respectively. The time and space performance of these two programs are evaluated in comparison with eSAIS that is also implemented using STXXL. In our experiment, eSAIS runs the fastest for the input strings not larger than 16 GiB, but it is slower than SAIS-PQ+ for the only two input strings of 32 and 48.44 GiB. For the average peak disk requirements, eSAIS and SAIS-PQ+ are around 23n and 15n bytes, respectively.

关键词： Suffix array sorting algorithm External memory

来源：评论

学校读者我要写书评

暂无评论

Rule-Based Classification Models of Molecular Autofluorescence

引用

JOURNAL OF CHEMICAL INFORMATION AND MODELING 2015年第2期55卷 434-445页

作者： Su, Bo-Han Tu, Yi-Shu Lin, Olivia A. Harn, Yeu-Chern Shen, Meng-Yu Tseng, Yufeng J. Natl Taiwan Univ Dept Comp Sci & Informat Engn Taipei 106 Taiwan Natl Taiwan Univ Grad Inst Biomed Elect & Bioinformat Taipei 106 Taiwan Natl Taiwan Univ Sch Med Drug Res Ctr Taipei 106 Taiwan

Fluorescence-based detection has been commonly used in high-throughput screening (HTS) assays. Autofluorescent compounds, which can emit light in the absence of artificial fluorescent markers, often interfere with the detection of fluorophores and result in false positive signals in these assays. This interference presents a major issue in fluorescence-based screening techniques. In an effort to reduce the time and cost that will be spent on prescreening of autofluorescent compounds, in silico autofluorescence prediction models were developed for selected fluorescence-based assays in this study. Five prediction models were developed based on the respective fluorophores used in these HTS assays, which absorb and emit light at specific wavelengths (excitation/emission): Alexa Fluor 350 (A350) (340 nm/450 nm), 7-amino-4-trifluoromethyl-coumarin (AFC) (405 nm/520 nm), Alexa Fluor 488 (A488) (480 nm/540 nm), Rhodamine (547 nm/598 nm), and Texas Red (547 nm/618 nm). The C5.0 rule-based classification algorithm and PubChem 2D chemical structure fingerprints were used to develop prediction models. To optimize the accuracies of these prediction models despite the highly imbalanced ratio of fluorescent versus nonfluorescent compounds presented in the collected data sets, oversampling and undersampling strategies were applied. The average final accuracy achieved for the training set was 97%, and that for the testing set was 92%. In addition, five external data sets were used to further validate the models. Ultimately, 14 representative structural features (or rules) were determined to efficiently predict autofluorescence in data sets containing both fluorescent and nonfluorescent compounds. Several cases were illustrated in this study to demonstrate the applicability of these rules.

关键词： sorting algorithm Newton meter fluor Fluorophores Texas red Rule Data Collection Methods Dataset Prediction models Alexa Fluorescence prescreening TEST SETS Oversampling autofluorescence

来源：评论

学校读者我要写书评

暂无评论

Ultra-fast electromagnetic transient model of the modular multilevel converter for HVDC studies

Ultra-fast electromagnetic transient model of the modular mu...

引用

12th IET International Conference on AC and DC Power Transmission (ACDC 2016)

作者： Jianzhong Xu Hui Ding Shengtao Fan Aniruddha M. Gole Chengyong Zhao Yan Xiong North China Electric Power University China University of Manitoba Canada North China Electric Power University - Baoding Campus Baoding Hebei CN

This paper introduces a very fast electromagnetic transient (EMT) simulation model for the HVDC modular multi-level converter (MMC) that maintains the identity of each switching level, but achieves computation speeds comparable to much simplified averaged models. In contrast with previously developed fast MMC simulation models, this model achieves its speedup by using a special time-varying Thévenin equivalent based algorithm that uses the A-stable Backward Euler (BE) integration method. The approach assumes the switch's “ON” condition as a conduction resistance, whereas the “OFF” condition is modeled by an ideal open circuit. The approach was compared with a full detailed simulation (no simplifications), and showed negligible loss in accuracy and yet had a much reduced computational time, comparable with the much less accurate averaged-value models (AVM).

关键词： sorting algorithm Backward Euler (BE) carrier phase-shifted sinusoidal pulse-width modulation (CPS-SPWM) modular multilevel converter (MMC) nearest level control (NLC) sub-module (SM)

来源：评论

学校读者我要写书评

暂无评论

CONVOLUTIONAL SPARSE CODING CLASSIFICATION MODEL FOR IMAGE CLASSIFICATION

CONVOLUTIONAL SPARSE CODING CLASSIFICATION MODEL FOR IMAGE C...

引用

IEEE International Conference on Image Processing

作者： Boheng Chen Jie Li Biyun Ma Gang Wei School of Electronic and Information South China University of Technology

ISBN: (纸本)9781467399623

In this paper, we present a novel classification model which combines the convolutional sparse coding framework with the classification strategy. In the training phase, the proposed model trained a convolutional filter bank by all images of each class. In the test phase, the label of test image is determined by all convolutional filter banks. Compared with canonical sparse representation and dictionary learning classification algorithm, more representative information of the corresponding images could be captured by the trained filters, thus better classification performance can be obtained. Experimental results on some image benchmark databases demonstrated the effectiveness of the proposed method.

关键词： Convolutional sparse coding Sparse representation Convolutional filter Image representation Image classification Image classification sparse representation image representation Taxonomy images Filter banks sorting algorithm

来源：评论

学校读者我要写书评

暂无评论

A Matlab/Simulink model for capacitor voltages balancing in modular multilevel converters

A Matlab/Simulink model for capacitor voltages balancing in ...

引用

8th IET International Conference on Power Electronics, Machines and Drives (PEMD 2016)

作者： Nada Ahmed Ahmed Massoud Qatar University Doha Ad Dawhah QA

ISBN: (纸本)9781510825666

The modular multilevel converter (MMC) is a potential candidate for high power applications due to its modularity, availability, and enhanced power quality features. Nevertheless, the voltage balancing of the floating capacitors in the series-connected submodules (SMs) of the MMC represents a bottleneck in this type of multilevel converters. In this paper, a generalized voltage balancing technique for MMCs is modelled using Matlab/Simulink platform. The presented model can be simply extended to any number of levels where the switching frequency can be as low as the fundamental frequency. A 9-level MMC is modelled and simulated using half bridge topology where the switches are controlled using phase disposition carrier-based pulse width modulation (PD-PWM) scheme to validate the proposed concept.

关键词： MMC sorting algorithm Voltage balancing Fundamental switching frequency PD-PWM voltage balance METAL MATRIX COMPOSITES Mitocin-C sorting algorithm Modular multilevel converters multistage converter Capacitors Power quality

来源：评论

学校读者我要写书评

暂无评论

Region Ranking SVM for Image Classification

Region Ranking SVM for Image Classification

引用

IEEE Conference on Computer Vision and Pattern Recognition

作者： Zijun Wei Minh Hoai Stony Brook University

ISBN: (纸本)9781467388528

The success of an image classification algorithm largely depends on how it incorporates local information in the global decision. Popular approaches such as average-pooling and max-pooling are suboptimal in many situations. In this paper we propose Region Ranking SVM (RRSVM), a novel method for pooling local information from multiple regions. RRSVM exploits the correlation of local regions in an image, and it jointly learns a region evaluation function and a scheme for integrating multiple regions. Experiments on PASCAL VOC 2007, VOC 2012, and ILSVRC2014 datasets show that RRSVM outperforms the methods that use the same feature type and extract features from the same set of local regions. RRSVM achieves similar to or better than the state-of-the-art performance on all datasets.

关键词： Image classification local information. Support Vector Network Dataset Volatile Organic Compounds sorting algorithm Pascal Imagery (Psychotherapy) images

来源：评论

学校读者我要写书评

暂无评论

OBJECT CLASSIFICATION IN INFRARED IMAGES USING DEEP REPRESENTATIONS

OBJECT CLASSIFICATION IN INFRARED IMAGES USING DEEP REPRESEN...

引用

IEEE International Conference on Image Processing

作者： Erhan Gundogdu Aykut Koc A. Aydin Alatan Intelligent Data Analytics Research Program Dept. Aselsan Research Center Center for Image Analysis (OGAM) METU Ankara

ISBN: (纸本)9781467399623

In this study, we address the problem of infrared (IR) object classification that divides the object appearance space hierarchically with a binary decision tree structure. Binary decisions are made by using the special features of the object appearances. These features are extracted using a fully connected deep neural network learnt by training samples. At each node of the tree, we train individual deep CNNs such that each node specializes in its corresponding subspace. The proposed classification algorithm is evaluated in our generated dataset, which consists of IR targets collected from different video records obtained from different IR sensors (both midwave and longwave) and taken from real world field. The generated dataset consists of four different class labels as ship/boat, tank, plane and helicopter containing a total of 16K samples. Using the proposed tree-based classifier, we observe a favourable performance increase in our dataset against a single deep CNN classifier.

关键词： Infrared Classification Thermal targets Thermal dataset generation Tree-based classification Taxonomy Infrared radiation sorting algorithm Dataset Binary Decision Tree tree sort Infrared sensors Video Recording Helicopters representations Line of nodes Classifiers binary decision

来源：评论

学校读者我要写书评

暂无评论

A Survey of Face Recognition Techniques and Comparative Study of Various Bi-Modal and Multi-Modal Techniques

A Survey of Face Recognition Techniques and Comparative Stud...

引用

IEEE International Conference on Industrial and Information Systems

作者： Anand Handa Rashi Agarwal Narendra Kohli Department of CSE PSIT COE Kanpur (India) Department of IT UIET CSJMU Kanpur (India) Department of CSE HBTIU Kanpur (India)

Face Recognition is one of the most demanding problems in computer vision and image processing. Face recognition techniques can be divided into following categories: methods which are applicable on intensified images;images derived from video sequences;and images that require three-dimensional information and data. In this paper, we have compared the various Bi-Modal and Multi-Modal techniques. Speech recognition is an essential component for various applications such as building interfaces for natural human machines. Acoustic speech is used as the only input for various speech based automatic recognition system. Multimodal recognition is as an indigenous component of the next level speech-based systems. The main objective of this review paper is to compare the various components of bimodal recognition and it aims at some important ongoing research issues in the field of image and face recognition.

关键词： Hidden Markov models Speech recognition Face recognition Speech Face Classification algorithms Feature extraction Facial Recognition Face recognition techniques sorting algorithm Speech recognition images Hidden Markov models Speech Feature extraction Face computer vision

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：