一适用于超多类手写汉字识别的新改型Adaboost算法

提出一种适用于超多类手写汉字识别的新改型Adaboost算法,采用基于描述性模型的多类分类器（modified quadratic discriminant function,MQDF）作为Adaboost基元分类器,可直接进行多类分类,无需将多类问题转化为多个两类问题处理,其训练复杂度大大低于已有的多类Adaboost算法。算法提出根据广义置信度更新样本权重,实验证明这种算法适用于大规模多类分类问题。为了降低算法的识别复杂度,提出从所有训练后得到的Adaboost基元分类器组中选择一个最优的基元分类器作为最终分类器的方法进行删减。在HCL2000及THOCR-HCD数据集上进行实验证明,所提改型Adaboost算法提高了识别率的有效性,该算法的相对错误率比现有最优算法分别下降了14.3 %,8.1 %和19.5 %。     

用于手写汉字识别的改型Adaboost算法

针对大部分多类Adaboost算法因训练复杂度过高而难以应用于手写汉字识别这种大类别数分类的问题,提出了一种新的改型的多类Adaboost算法.该算法采用基于描述性模型的多类分类器--改进的二次鉴别函数(MQDF)分类器作为基元分类器,可直接进行多类分类,无需将多类问题转化为多个两类问题处理,大大降低了训练复杂度.此外,该算法根据广义置信度更新样本权重,实验证明此方法简单有效.为了降低算法的识别复杂度,对训练后得到的基元分类器组进行删减,仅保留一个最优的基元分类器作为最终分类器.在HCL2000及THOCR-HCD数据集上进行的实验表明,该算法的相对错误率比现有算法分别下降了14.3%、8.1%和19.5%.     

基于共形几何代数与二次规划的分类器设计

提出一种基于共形几何代数与二次规划的分类器设计方法.从新的角度出发,讨论了运用共形几何代数理论来构造最优分类超球可分问题的可行性和简便性,首先介绍了基于共形几何代数的分类超球面的几何表示,并用此表示将二类最优分类超球面的可分问题转化二次规划的训练学习问题,在此基础上分析了多类分类器的设计和训练方法.该算法保留了最大分类间隔理论的优点,将二类最优平面可分推广到最优超球可分,简化了其运算复杂度,仿真实验表明,该学习算法简洁明确,对于算法的集成,提高效率有着很重要的意义.     

基于Adaboost算法选取和组合SVM的行人检测算法

行人检测是计算机视觉中一个重要的研究方向,为了提高行人的识别精度,将支持向量机(Sup-port Vector Machine,SVM)和Adaboost算法结合起来,SVM是基于结构风险最小化准则的新型机器学习算法,适合小样本学习并且能够有效地抑制过拟合问题,Adaboost基于最小化训练错误率,一般使用易训练的分类器作为弱分类器.由于SVM比较难训练,因此将样本集划分形成多个训练集,然后利用正样本和不同的负样本组成不同训练集反复训练,最后通过Adaboost对训练集生成的SVM模型筛选出具有最小错误率的SVM分类器并且采用投票机制形成最终的强分类器.实验结果表明,在FPPW(false positive per window)为10-5时检测率能够达到30%,检测效果优于单个SVM算法训练出来的分类器模型,用行人测试库测试,该方法取得了较好的检测效果并且具有较强的鲁棒性.     

基于人工鱼群的Gentle Adaboost快速训练算法

Gentle Adaboost算法训练弱分类器时,需要遍历特征空间,将分类结果最好的特征作为弱分类器,这将消耗大量的时间。本文提出了一种基于人工鱼群的Gentle Adaboost快速训练算法。人工鱼群算法能够模拟鱼群行为策略,有效的对特征空间快速搜索,减少需要计算的特征数,缩短训练时间。在保证检测效果的条件下,通过对MIT和FERET人脸数据库部分样本的训练,新方法的训练时间约能缩短至原始训练时间的1/4。     

基于遗传算法优化支持向量机的超声图像缺陷分类

超声图像缺陷在分类时由于存在样本数量少、样本类别多、不易区分等问题,分类的准确率较低。针对这些问题,提出了基于遗传算法优化支持向量机的超声图像缺陷分类方法。该方法首先通过图像处理提取超声图像缺陷的特征数据,然后训练支持向量机作为超声图像缺陷分类器,最后采用遗传算法优化参数求得最优的分类器。实验结果表明,提出的超声图像缺陷分类器在识别率方面优于其他方法的分类器,综合识别率达到了90%,可以有效地辅助工作人员对超声图像缺陷进行分类识别。     

基于AlexNet-Adaboost的多工况滚动轴承故障识别方法

针对实际工程中滚动轴承多工况下传统故障诊断方法识别率偏低的情况.提出了一种基于AlexNet-Adaboost相结合的滚动轴承故障识别方法.以滚动轴承信号的时频图作为模型输入、分类结果作为模型输出,训练多个AlexNet基分类器;在此基础上利用Adaboost(自适应提升)算法进一步提升得到强分类器,将多工况下滚动轴承...     

基于遗传算法进化的数字图像处理

人脸表情识别是目前数字图像处理领域比较活跃的研究课题。本文提出一种采用遗传算法进化的支持向量机对人脸表情进行分类的新型算法。先提取静态人脸表情特征,然后采用遗传算法自动选择最优的支持向量机核函数,最后采用这种新型分类器进行了人脸表情的分类和识别。在Yale人脸表情库上进行了测试人不参与训练的仿真实验,并与最近邻分类器进行比较,提出的方法取得了更好的识别结果。     

自适应免疫算法的SVME用于水下目标识别

针对用支持向量机集成提高水下目标识别正确率会使识别系统更加复杂的问题,提出了一种以自适应免疫算法(AIA)的支持向量机选择性集成(SVME)算法(即AIA-SVME算法)进行分类器优化选择,对实测水下目标声信号进行分类识别.与分类器全部集成的识别实验对比证明,该算法在选择9%的分类器后仍可以达到分类器全部集成的识别效果,不仅保证了识别精度,还使得识别系统大幅度精简,节省在线识别的时间.该研究对于水下目标分类决策优化集成的新方法探索具有重要理论价值和实际意义.     

基于改进差分进化算法的运动想象脑机接口频带选择

由于人脑对事件响应频带各不相同,为了准确确定个体最优滤波频带,提出一种多策略变异算子和时变非线性交叉因子差分进化算法对运动想象EEG频带进行处理,采用共空间模式算法提取特征向量,利用线性分类器进行分类识别。使用该方案对BCI competition III-dataset 4a受试者的EEG数据进行了10次5倍交叉分类实验。实验结果表明,该算法稳定性强、耗时少,解决了运动想象BCI特征提取中的最优频带选择问题。     

Multi-class differentiation of cannabis seedlings in a forensic context

This article presents an experimental study about the classification ability of several classifiers for multi-class classification of cannabis seedlings. As the cultivation of drug type cannabis is forbidden in Switzerland law enforcement authorities regularly ask forensic laboratories to determinate the chemotype of a seized cannabis plant and then to conclude if the plantation is legal or not. This classification is mainly performed when the plant is mature as required by the EU official protocol and then the classification of cannabis seedlings is a time consuming and costly procedure. A previous study made by the authors has investigated this problematic [1] and showed that it is possible to differentiate between drug type (illegal) and fibre type (legal) cannabis at an early stage of growth using gas chromatography interfaced with mass spectrometry (GC-MS) based on the relative proportions of eight major leaf compounds. The aims of the present work are on one hand to continue former work and to optimize the methodology for the discrimination of drug- and fibre type cannabis developed in the previous study and on the other hand to investigate the possibility to predict illegal cannabis varieties. Seven classifiers for differentiating between cannabis seedlings are evaluated in this paper, namely Linear Discriminant Analysis (LDA), Partial Least Squares Discriminant Analysis (PLS-DA), Nearest Neighbour Classification (NNC), Learning Vector Quantization (LVQ), Radial Basis Function Support Vector Machines (RBF SVMs), Random Forest (RF) and Artificial Neural Networks (ANN). The performance of each method was assessed using the same analytical dataset that consists of 861 samples split into drug- and fibre type cannabis with drug type cannabis being made up of 12 varieties (i.e. 12 classes). The results show that linear classifiers are not able to manage the distribution of classes in which some overlap areas exist for both classification problems. Unlike linear classifiers, NNC and RBF SVMs best differentiate cannabis samples both for 2-class and 12-class classifications with average classification results up to 99% and 98%, respectively. Furthermore, RBF SVMs correctly classified into drug type cannabis the independent validation set, which consists of cannabis plants coming from police seizures. In forensic case work this study shows that the discrimination between cannabis samples at an early stage of growth is possible with fairly high classification performance for discriminating between cannabis chemotypes or between drug type cannabis varieties.     

基于独立特征选择与相关向量机的变载荷轴承故障诊断

在线状态监控与故障诊断具有很大的经济与安全意义,提出了一种基于独立特征选择(IFS)与相关向量机(RVM)的智能故障诊断模型用于变载荷条件下识别多类轴承故障及其故障程度。首先混合空载(0hp)与满载(3hp)两种载荷状态下的实验数据作为训练样本;其次提取时域统计特征与全小波包域节点能量特征作为候选特征;接着采用一种改进的Fisher特征选择方法为每两类故障状态独立选择具有最大分类能力的最优特征子集;然后用“一对一”的方法训练多个RVM二类子分类器;最后采用“最大概率赢”的策略组合所有子分类器构成IFS_RVM多类故障诊断模型。用未知载荷(1hp,2hp)下的实验数据验证了模型的有效性,得到99.58%的极高诊断精度,实验结果表明,该模型精度高、鲁棒性强,满足变载荷条件下在线故障诊断的需要     

Support vector machine ensembles for breast cancer type prediction from mid-FTIR micro-calcification spectra

Over the past years Fourier transform infrared (FTIR) spectroscopy has been demonstrated as a prospective tool for cancer diagnostics. In order to apply FTIR spectroscopy as a routine tool for biomedical diagnostics of tissue samples, strong and reliable classifiers are needed. Frequently, the number of available tissue samples is restricted and due to that data sets consist of a small number of samples, often less than 100. This can result in unstable classifiers, which perform poorly on unseen data. In this work we present a way to overcome this limitation by aggregating several support vector machines in to an ensemble. Different ensemble systems, including bagging, boosting and tree-based models, were investigated for a FTIR data set acquired from different types and stages of breast cancer. It was found that an ensemble system predicts 88.9% of the unseen multi-class test set correctly. In comparison a single classifier only achieved a predictive performance of 66.7%. As these results show, the application of SVM ensembles in biomedical diagnostics using FTIR spectroscopy can be highly beneficial.     

基于时间序列数据和支持向量机的纳米加工AFM刀尖损伤监测

研究了基于支持向量机(SVM)的时间序列数据分析和模式识别,以监测基于AFM尖端的纳米加工过程在加工性能和尖端磨损方面的状态变化。具有瞬态、非线性和非静止特性的时间序列数据(即来自过程的加工力)由数据采集系统收集。提取3种状态检测特征,包括最大侧向加工力、侧向加工力值峰间距以及侧向加工力的方差,以对纳米加工过程的状态进行分类。构造具有(高斯)径向基核函数(RBF内核)的定向非循环图支持向量机(DAGSVM)以识别尖端状态。使用多元SVM分类机,将加工过程和刀尖磨损分为初始磨损、过渡区域磨损以及尖端失效(破裂/磨损严重的加工/不加工)3个区域。实验数据表明,二元和三元分类中SVM的准确率均超过94.73％。     

A multi-view method of scientific paper classification via heterogeneous graph embeddings

The classification task of scientific papers can be implemented based on contents or citations. In order to improve the performance on this task, we express papers as nodes and integrate scientific papers’ contents and citations into a heterogeneous graph. It has two types of edges. One type represents the semantic similarity between papers, derived from papers’ titles and abstracts. The other type represents the citation relationship between papers and the journals or proceedings of conferences of their references. We utilize a contrastive learning method to embed the nodes in the heterogeneous graph into a vector space. Then, we feed the paper node vectors into classifiers, such as the decision tree, multilayer perceptron, and so on. We conduct experiments on three datasets of scientific papers: the Microsoft Academic Graph with 63,211 scientific papers in 20 classes, the Proceedings of the National Academy of Sciences with 38,243 scientific papers in 18 classes, and the American Physical Society with 443,845 scientific papers in 5 classes. The experimental results on the multi-class task show that our multi-view method scores the classification accuracy up to 98%, outperforming state-of-the-arts.

     

A Novel Combinational Convolutional Neural Network for Automatic Food-Ingredient Classification

With the development of deep learning and Convolutional Neural Networks (CNNs), the accuracy of automatic food recognition based on visual data have significantly improved. Some research studies have shown that the deeper the model is, the higher the accuracy is. However, very deep neural networks would be affected by the overfitting problem and also consume huge computing resources. In this paper, a new classification scheme is proposed for automatic food-ingredient recognition based on deep learning. We construct an up-to-date combinational convolutional neural network (CBNet) with a subnet merging technique. Firstly, two different neural networks are utilized for learning interested features. Then, a well-designed feature fusion component aggregates the features from subnetworks, further extracting richer and more precise features for image classification. In order to learn more complementary features, the corresponding fusion strategies are also proposed, including auxiliary classifiers and hyperparameters setting. Finally, CBNet based on the well-known VGGNet, ResNet and DenseNet is evaluated on a dataset including 41 major categories of food ingredients and 100 images for each category. Theoretical analysis and experimental results demonstrate that CBNet achieves promising accuracy for multi-class classification and improves the performance of convolutional neural networks.     

Multi-class queueing models for performance analysis of computer systems

Queueing models, networks of queues in particular, have been found especially useful for estimating the performance of computer systems. Networks of queues with multiple customer classes provide a flexible framework for modelling computer systems, where a rich set of analytical results and techniques are available. When because of the complexity of the system being modelled the analytical results cannot be applied directly, they often point to fairly accurate approximation schemes. In this paper, we present a brief survey of some of the important results and techniques from the theory of multi-class queueing networks. We also present a case study to illustrate how these results and techniques are used in a real-life situation where many of the modelling constraints are violated.     

基于自适应粒子群算法和支持向量机的控制图模式识别

针对目前多品种、复杂化的生产趋势,提出了一种基于自适应变异的粒子群算法（AMPSO）和支持向量机（SVM）的控制图失效模式识别的方法。利用SVM小样本学习能力,设计一对一的SVM多分类器进行控制图模式识别,并利用AMPSO算法优化SVM核函数的参数。通过对10种控制图模式(6种基本模式和4种混合模式)的20维特征仿真数据对该方法进行检验,并通过与BP、SVM、PSO SVM识别方法的对比分析。仿真试验表明该方法有效提高了控制图模式的识别精度,达到9814%,而BP仅有75%,为控制图在线实时识别提供了一种可行的途径。 