Description: 基于多分类器组合的笔迹验证 －－文章-Based on the composition of the classifier based on the handwriting test multiple classifiers combination of handwriting authentication -- article Platform: |
Size: 216233 |
Author:Yuan |
Hits:
Description: Boosting is a meta-learning approach that aims at combining an ensemble of weak classifiers to form a strong classifier. Adaptive Boosting (Adaboost) implements this idea as a greedy search for a linear combination of classifiers by overweighting the examples that are misclassified by each classifier. icsiboost implements Adaboost over stumps (one-level decision trees) on discrete and continuous attributes (words and real values). See http://en.wikipedia.org/wiki/AdaBoost and the papers by Y. Freund and R. Schapire for more details [1]. This approach is one of most efficient and simple to combine continuous and nominal values. Our implementation is aimed at allowing training from millions of examples by hundreds of features in a reasonable time/memory.
Platform: |
Size: 116681 |
Author:njustyw |
Hits:
Description: 基于多分类器组合的笔迹验证 －－文章-Based on the composition of the classifier based on the handwriting test multiple classifiers combination of handwriting authentication-- article Platform: |
Size: 216064 |
Author:Yuan |
Hits:
Description: 基于模板匹配与支持矢量机的人脸检测
梁路宏 艾海舟 肖习攀 叶航军 徐光佑 张钹
。本文提出了一种将模板匹配与支持矢量机（SVM）相结合的人脸检测算法。算法首先使用双眼—人脸模板对进行粗筛选，然后使用SVM分类器进行分类。在模板匹配限定的子空间内采用“自举”方法收集“非人脸”样本训练SVM，有效地降低了训练的难度。实验结果的对比数据表明，该算法是十分有效的。-Based on template matching and support vector machine Face Detection Liang Wang Ai Haizhou Road, Xi Pan Xiao Ye Zhang Bo Xu Guangyou military aircraft. This paper proposes a template matching and support vector machine (SVM) a combination of face detection algorithm. Algorithm the first to use both eyes- face template to carry out coarse filter, and then use the SVM classifier to classify. Template matching in the sub-space limit the use of Platform: |
Size: 1135616 |
Author:cy |
Hits:
Description: Boosting is a meta-learning approach that aims at combining an ensemble of weak classifiers to form a strong classifier. Adaptive Boosting (Adaboost) implements this idea as a greedy search for a linear combination of classifiers by overweighting the examples that are misclassified by each classifier. icsiboost implements Adaboost over stumps (one-level decision trees) on discrete and continuous attributes (words and real values). See http://en.wikipedia.org/wiki/AdaBoost and the papers by Y. Freund and R. Schapire for more details [1]. This approach is one of most efficient and simple to combine continuous and nominal values. Our implementation is aimed at allowing training from millions of examples by hundreds of features in a reasonable time/memory.
Platform: |
Size: 116736 |
Author:njustyw |
Hits:
Description: VM分类器通常具有较高的分类精度。我这里不想过多的去说SVM是怎么回事，只是提供一种使用SVM进行判别的方法。决策树与SVM的结合，可以分多类。-VM classifier usually has a higher classification accuracy. I do not want too much here to say how the matter SVM, SVM is used to provide a method for identification. The combination of decision tree and SVM, can be divided into many types. Platform: |
Size: 105472 |
Author:steve |
Hits:
Description: ROC curves illustrate performance on a binary classification problem where classification is based on simply thresholding a set of scores at varying levels. Lenient thresholds give high sensitivity but low specificity, strict thresholds give high specificity but low sensitivity the ROC curve plots this trade-off over a range of thresholds (usually with sens vs 1-spec, but I prefer sens vs spec this code gives you the option).
It is theoretically possible to operate anywhere on the convex hull of an ROC curve, so this is plotted too. The area under the curve (AUC) for a ROC plot is a measure of overall accuracy, and the area under the ROCCH is a kind of upper bound on what might be achievable with a weighted combination of differently thresholded results from the given classifier
-ROC curves illustrate performance on a binary classification problem where classification is based on simply thresholding a set of scores at varying levels. Lenient thresholds give high sensitivity but low specificity, strict thresholds give high specificity but low sensitivity the ROC curve plots this trade-off over a range of thresholds (usually with sens vs 1-spec, but I prefer sens vs spec this code gives you the option).
It is theoretically possible to operate anywhere on the convex hull of an ROC curve, so this is plotted too. The area under the curve (AUC) for a ROC plot is a measure of overall accuracy, and the area under the ROCCH is a kind of upper bound on what might be achievable with a weighted combination of differently thresholded results from the given classifier
Platform: |
Size: 4096 |
Author:saadat |
Hits:
Description: 基于奇异值分解的人脸识别方法
梁毅雄 龚卫国 潘英俊 李伟红 刘嘉敏 张红梅
提出了一种将傅里叶变换和奇异值分解相结合的人脸自动识别方法.首先对人脸图像进行傅里叶变换,得到其具有位移不变特性的振幅谱表征.其次,从所有训练图像样本的振幅谱表征中给定标准脸并对其进行奇异值分解,求出标准特征矩阵,再将人脸的振幅谱表征投影到标准特征矩阵后得到的投影系数作为该人脸的模式特征.然后,对经典的最近邻分类器算法进行了改进,并采用模式特征之间的欧式距离作为相似性度量,从而完成对未知人脸的识别.采用ORL (Olivetti Research Laboratory)人脸库对本文提出的人脸识别方法进行验证,获得了100.00 的识别率.实验结果表明,本方法优于现有的基于奇异值分解的人脸识别方法,且对表情、姿态变换等具有一定的鲁棒性.
-Face recognition based on singular value decomposition method
Deliberate simultaneously Gong Weiguo Li Wei Hung Stephen Lau, Hong-Mei Zhang Ying-Jun Pan
Paper, a Fourier transform and singular value decomposition of the combination of automatic face recognition. First of all, the face image by Fourier transformation, it has the same characteristics of the displacement amplitude spectra. Secondly, all training The amplitude spectrum of the sample images given in standard face representation and its singular value decomposition, find the standard characteristic matrix, then the amplitude of spectral characterization of human faces projected onto the standard characteristic matrix of projection coefficients obtained as the face of the model features . Then, the classical nearest neighbor classifier is improved, and the use of Euclidean distance between pattern features as the similarity measure, thus completing the identification of unknown human faces. using ORL (Olivetti Research La Platform: |
Size: 58368 |
Author:houhj |
Hits:
Description: A Novel Efficient Approach for Audio Segmentation
a novel approach to audio segmentation
is presented. The problem of detecting audio segments’
limits is treated as a binary classification task.
Frames are classified as “segment limits” vs “nonsegment
limits”. For each audio frame a spectrogram
is computed and eight feature values are extracted from
respective frequency bands. Final decisions are taken
based on a classifier combination scheme Platform: |
Size: 225280 |
Author:kvga |
Hits:
Description: AdaBoost 是一种将弱分类器线性结合成强分类器的方法，由Jiri Matas 和 Jan Sˇ ochman提出-AdaBoost is a linear combination of the weak classifiers into a strong classifier method, by Jiri Matas and Jan S ochman proposed Platform: |
Size: 3072 |
Author:董晶晶 |
Hits:
Description: 结合DCT和BP神经网络进行人脸识别。先利用DCT提取特征，然后利用BP神经网络分类，在ORL人脸库上测试效果不错。-The combination of DCT and BP neural network for face recognition. First DCT Feature Extraction, and then use the BP neural network classifier, a good test results on the ORL face database. Platform: |
Size: 2048 |
Author:尹贺峰 |
Hits:
Description: VM分类器通常具有较高的分类精度。我这里不想过多的去说SVM是怎么回事，，只是提供一种使用SVM进行判别的方法。决策树与SVM的结合，可以分多类。
-VM classifier usually has high classification accuracy. I do not want too much to say that SVM is how, just a SVM is used to discriminate. Combination of decision tree and SVM multi-class. Platform: |
Size: 106496 |
Author:获得 |
Hits:
Description: 基于纹理分析笔迹鉴别系统的设计与实现,文中从笔迹图像预处理、特征提取、分类器以及分类器组合等方而展开研究，设计和实现了一个基于文本独立的离线手写体笔迹鉴别系统软件.-Design and Implementation of the writer identification system based on texture analysis, the paper from the handwriting image preprocessing, feature extraction, classifiers, and classifier combination study, design and implementation of a separate text-based Offline Handwriting identification system software. Platform: |
Size: 972800 |
Author:Chris |
Hits:
Description: LinearClassifier,线性分类器进行分类，通过制作一个基于特征的线性组合的价值分类决策。对象的特征也被称为特征值，通常是一个向量提出的机器称为一个特征向量。-LinearClassifier, linear classifier for classification, through the production of a feature-based value of a linear combination of classification decisions. Characteristic of the object is also called the characteristic value, usually a vector proposed machine called a feature vector. Platform: |
Size: 5120 |
Author:lc |
Hits:
Description: 随机森林是由多棵树组成的分类或回归方法。主要思想来源于Bagging算法，Bagging技术思想主要是给定一弱分类器及训练集，让该学习算法训练多轮，每轮的训练集由原始训练集中有放回的随机抽取，大小一般跟原始训练集相当，这样依次训练多个弱分类器，最终的分类由这些弱分类器组合，对于分类问题一般采用多数投票法，对于回归问题一般采用简单平均法。随机森林在bagging的基础上，每个弱分类器都是决策树，决策树的生成过程中中，在属性的选择上增加了依一定概率选择属性，在这些属性中选择最佳属性及分割点，传统做法一般是全部属性中去选择最佳属性，这样随机森林有了样本选择的随机性，属性选择的随机性，这样一来增加了每个分类器的差异性、不稳定性及一定程度上避免每个分类器的过拟合（一般决策树有过拟合现象），由此组合分类器增加了最终的泛化能力。-Random Forest classification or regression trees by the multi-component. The main idea comes the Bagging algorithm, Bagging technology thinking mainly given a set of weak classifiers and training, so that the learning algorithm to train several rounds, each round of training set by the original training set is randomly selected with replacement, with the original size of the general training set fairly, and in turn train a plurality of weak classifiers, the final classification by the weak classifier combination of these, for the general classification of a majority voting method, commonly used for regression simple average method. Random forests bagging on the basis of each of the weak classifiers are decision trees, decision tree in the generation process, the choice of property on the increase in the probability of selection according to certain attributes, and choose the best attributes of these properties in the split point traditional practice is generally to choose the best of Platform: |
Size: 1024 |
Author:小代 |
Hits:
Description: 《智能web算法》涵盖了五类重要的智能算法：搜索、推荐、聚类、分类和分类器组合，并结合具体的案例讨论了它们在Web应用中的角色及要注意的问题。除了第1章的概要性介绍以及第7章对所有技术的整合应用外，第2~6章以代码示例的形式分别对这五类算法进行了介绍。- Intelligent web algorithm covers five important intelligent algorithms: search, recommendation, clustering, classification and classifier combination, combined with specific cases to discuss their role in Web applications and pay attention to the problem. In addition to Chapter 1 of the brief introduction and Chapter 7 of the integration of all technology applications, the first two to six chapters in the form of code examples, respectively, these five algorithms were introduced. Platform: |
Size: 4684800 |
Author:LeoLie |
Hits:
Description: 粒子群优化算起源于对鸟群、鱼群以及对某些社会行为的模拟，是一种基于群体智能的进化计算技术。而小生境技术则起源于遗传算法，这种方法能使基于群体的随机优化算法形成物种，从而使相应的优化算法具有发现多个最优解的能力。而多分类器集成技术则是通过多个分类器进行某种组合来决定最终的分类，以取得比单个分类器更好的性能。多分类器集成技术要求基元分类器不仅个体性能要好并且其差异度要大，这与小生境技术形成物种的能力具有很多内在的相似性。目前己经有研究者将小生境技术应用于多分类器集成，但由于传统的小生境技术仍然不完善，存在一些内在的陷，因而这些应用还不成熟和完善。
(Particle swarm optimization (partieleSwarmOptimization) originated in the birds, fish, and of a
Some simulation of social behavior, is a swarm intelligence-based evolutionary computing. The origin of the niche technology is In genetic algorithms, this method can make random optimization algorithm based on the formation of groups of species, so that the appropriate priority
Algorithm has the ability to find multiple optimal solutions. The integration technology of multiple classifiers is through multiple classifiers into
Some combination of the line to determine the final classification, in order to obtain better than a single classifier performance. Integration of multiple classifiers
Technical requirements for primitive classification is not only better individual performance and the difference to a large degree, which form a niche technology The ability of species has many inherent similarities. The researchers will now have a niche technology used in multisection
Class ens) Platform: |
Size: 5953536 |
Author:dreamer |
Hits:
Search - classifier combination