Combination of features through weighted ensembles for image classification

Image classification is a multi-class problem that is usually tackled with ensembles of binary classifiers. Furthermore, one of the most important challenges in this field is to find a set of highly discriminative image features for reaching a good performance in image classification. In this work we propose to use weighted ensembles as a method for feature combination. First, a set of binary classifiers are trained with a set of features and then, the scores are weighted with distances obtained from another set of feature vectors. We present two different approaches to weight the score vector: (1) directly multiplying each score by the weights and (2) fusing the scores values and the distances through a Neural Network. The experiments have shown that the proposed methodology improves classification accuracy of simple ensembles and even more it obtains similar classification accuracy than state-of-the-art methods, but using much less parameters.     

Deep Multimodal Fusion: A Hybrid Approach

We propose a novel hybrid model that exploits the strength of discriminative classifiers along with the representation power of generative models. Our focus is on detecting multimodal events in time varying sequences as well as generating missing data in any of the modalities. Discriminative classifiers have been shown to achieve higher performances than the corresponding generative likelihood-based classifiers. On the other hand, generative models learn a rich informative space which allows for data generation and joint feature representation that discriminative models lack. We propose a new model that jointly optimizes the representation space using a hybrid energy function. We employ a Restricted Boltzmann Machines (RBMs) based model to learn a shared representation across multiple modalities with time varying data. The Conditional RBMs (CRBMs) is an extension of the RBM model that takes into account short term temporal phenomena. The hybrid model involves augmenting CRBMs with a discriminative component for classification. For these purposes we propose a novel Multimodal Discriminative CRBMs (MMDCRBMs) model. First, we train the MMDCRBMs model using labeled data by training each modality, followed by training a fusion layer. Second, we exploit the generative capability of MMDCRBMs to activate the trained model so as to generate the lower-level data corresponding to the specific label that closely matches the actual input data. We evaluate our approach on ChaLearn dataset, audio-mocap, as well as the Tower Game dataset, mocap-mocap as well as three multimodal toy datasets. We report classification accuracy, generation accuracy, and localization accuracy and demonstrate its superiority compared to the state-of-the-art methods.     

Semisupervised learning for a hybrid generative/discriminative classifier based on the maximum entropy principle

This paper presents a method for designing semi-supervised classifiers trained on labeled and unlabeled samples. We focus on probabilistic semi-supervised classifier design for multi-class and single-labeled classification problems, and propose a hybrid approach that takes advantage of generative and discriminative approaches. In our approach, we first consider a generative model trained by using labeled samples and introduce a bias correction model, where these models belong to the same model family, but have different parameters. Then, we construct a hybrid classifier by combining these models based on the maximum entropy principle. To enable us to apply our hybrid approach to text classification problems, we employed naive Bayes models as the generative and bias correction models. Our experimental results for four text data sets confirmed that the generalization ability of our hybrid classifier was much improved by using a large number of unlabeled samples for training when there were too few labeled samples to obtain good performance. We also confirmed that our hybrid approach significantly outperformed generative and discriminative approaches when the performance of the generative and discriminative approaches was comparable. Moreover, we examined the performance of our hybrid classifier when the labeled and unlabeled data distributions were different.     

A Novel Computationally Efficient Approach to Identify Visually Interpretable Medical Conditions from 2D Skeletal Data

Timely identification and treatment of medical conditions could facilitate faster recovery and better health. Existing systems address this issue using custom-built sensors, which are invasive and difficult to generalize. A low-complexity scalable process is proposed to detect and identify medical conditions from 2D skeletal movements on video feed data. Minimal set of features relevant to distinguish medical conditions: AMF, PVF and GDF are derived from skeletal data on sampled frames across the entire action. The AMF (angular motion features) are derived to capture the angular motion of limbs during a specific action. The relative position of joints is represented by PVF (positional variation features). GDF (global displacement features) identifies the direction of overall skeletal movement. The discriminative capability of these features is illustrated by their variance across time for different actions. The classification of medical conditions is approached in two stages. In the first stage, a low-complexity binary LSTM classifier is trained to distinguish visual medical conditions from general human actions. As part of stage 2, a multi-class LSTM classifier is trained to identify the exact medical condition from a given set of visually interpretable medical conditions. The proposed features are extracted from the 2D skeletal data of NTU RGB + D and then used to train the binary and multi-class LSTM classifiers. The binary and multi-class classifiers observed average F1 scores of 77% and 73%, respectively, while the overall system produced an average F1 score of 69% and a weighted average F1 score of 80%. The multi-class classifier is found to utilize 10 to 100 times fewer parameters than existing 2D CNN-based models while producing similar levels of accuracy.     

基于分解策略处理多分类不均衡问题的方法

针对多分类不均衡问题,提出了一种新的基于一对一（one-versus-one,OVO）分解策略的方法。首先基于OVO分解策略将多分类不均衡问题分解成多个二值分类问题;再利用处理不均衡二值分类问题的算法建立二值分类器;接着利用SMOTE过抽样技术处理原始数据集;然后采用基于距离相对竞争力加权方法处理冗余分类器;最后通过加权投票法获得输出结果。在KEEL不均衡数据集上的大量实验结果表明,所提算法比其他经典方法具有显著的优势。     

一种新的多类支持向量机决策方法及其应用

讨论和比较了现有的几种多类SVM方法．在此基础上,提出了一种组合多个两类分类器结果的多类SVM决策方法．在该方法中,定义了新的决策函数,其值是在传统投票决策值的基础上乘以不同分类器的权重．新的多类SVM在一定程度上解决了传统投票决策方法的不可分区域问题,因此具有更好的分类性能．最后,将新方法作为关键技术应用于故障诊断实例,实际诊断结果证明了所提多类SVM决策方法的优越性．     

产生式与判别式线性混合分类器

产生式方法和判别式方法是解决分类问题的两种不同框架,具有各自的优势。为利用两种方法各自的优势,文中提出一种产生式与判别式线性混合分类模型,并设计一种基于遗传算法的产生式与判别式线性混合分类模型的学习算法。该算法将线性混合分类器混合参数的学习看作一个最优化问题,以两个基分类器对每个训练数据的后验概率值为数据依据,用遗传算法找出线性混合分类器混合参数的最优值。实验结果表明,在大多数数据集上,产生式与判别式线性混合分类器的分类准确率优于或近似于它的两个基分类器中的优者。     

Functional gradient ascent for Probit regression

This paper proposes two gradient based methods to fit a Probit regression model by maximizing the sample log-likelihood function. Using the property of the Hessian of the objective function, the first method performs weighted least square regression in each iteration of the Newton–Raphson framework, resulting in ProbitBoost, a boosting-like algorithm. Motivated by the gradient boosting algorithm [10], the second proposed approach maximizes the sample log-likelihood function by updating the fitted function a small step in the gradient direction, performing gradient ascent in functional space, resulting in Gradient ProbitBoost. We also generalize the algorithms to multi-class problems by two strategies, one of which is to use the gradient ascent to maximize the multi-class sample log-likelihood function for fitting all the classifiers simultaneously, and the second approach uses the one-versus-all scheme to reduce the multi-class problem to a series of binary classification problems. The proposed algorithms are tested on typical classification problems including face detection, cancer classification, and handwritten digit recognition. The results show that compared to the alternative methods, the proposed algorithms perform similar or better in terms of testing error rates.     

NMC: nearest matrix classification – A new combination model for pruning One-vs-One ensembles by transforming the aggregation problem

The One-vs-One strategy is among the most used techniques to deal with multi-class problems in Machine Learning. This way, any binary classifier can be used to address the original problem, since one classifier is learned for each possible pair of classes. As in every ensemble method, classifier combination becomes a vital step in the classification process. Even though many combination models have been developed in the literature, none of them have dealt with the possibility of reducing the number of generated classifiers after the training phase, i.e., ensemble pruning, since every classifier is supposed to be necessary.On this account, our objective in this paper is two-fold: (1) We propose a transformation of the aggregation step, which lead us to a new combination strategy where instances are classified on the basis of the similarities among score-matrices. (2) This fact allows us to introduce the possibility of reducing the number of binary classifiers without affecting the final accuracy. We will show that around 50% of classifiers can be removed (depending on the base learner and the specific problem) and that the confidence degrees obtained by these base classifiers have a strong influence on the improvement in the final accuracy.A thorough experimental study is carried out in order to show the behavior of the proposed approach in comparison with the state-of-the-art combination models in the One-vs-One strategy. Different classifiers from various Machine Learning paradigms are considered as base classifiers and the results obtained are contrasted with the proper statistical analysis.     

Hierarchical classification method based on selective learning of slacked hierarchy for activity recognition systems

Physical activity recognition using wearable sensors has gained significant interest from researchers working in the field of ambient intelligence and human behavior analysis. The problem of multi-class classification is an important issue in the applications which naturally has more than two classes. A well-known strategy to convert a multi-class classification problem into binary sub-problems is the error-correcting output coding (ECOC) method. Since existing methods use a single classifier with ECOC without considering the dependency among multiple classifiers, it often fails to generalize the performance and parameters in a real-life application, where different numbers of devices, sensors and sampling rates are used. To address this problem, we propose a unique hierarchical classification model based on the combination of two base binary classifiers using selective learning of slacked hierarchy and integrating the training of binary classifiers into a unified objective function. Our method maps the multi-class classification problem to multi-level classification. A multi-tier voting scheme has been introduced to provide a final classification label at each level of the solicited model. The proposed method is evaluated on two publicly available datasets and compared with independent base classifiers. Furthermore, it has also been tested on real-life sensor readings for 3 different subjects to recognize four activities i.e. Walking, Standing, Jogging and Sitting. The presented method uses same hierarchical levels and parameters to achieve better performance on all three datasets having different number of devices, sensors and sampling rates. The average accuracies on publicly available dataset and real-life sensor readings were recorded to be 95% and 85%, respectively. The experimental results validate the effectiveness and generality of the proposed method in terms of performance and parameters.     

An efficient orientation distance–based discriminative feature extraction method for multi-classification

Feature extraction is an important step before actual learning. Although many feature extraction methods have been proposed for clustering, classification and regression, very limited work has been done on multi-class classification problems. This paper proposes a novel feature extraction method, called orientation distance–based discriminative (ODD) feature extraction, particularly designed for multi-class classification problems. Our proposed method works in two steps. In the first step, we extend the Fisher Discriminant idea to determine an appropriate kernel function and map the input data with all classes into a feature space where the classes of the data are well separated. In the second step, we put forward two variants of ODD features, i.e., one-vs-all-based ODD and one-vs-one-based ODD features. We first construct hyper-plane (SVM) based on one-vs-all scheme or one-vs-one scheme in the feature space; we then extract one-vs-all-based or one-vs-one-based ODD features between a sample and each hyper-plane. These newly extracted ODD features are treated as the representative features and are thereafter used in the subsequent classification phase. Extensive experiments have been conducted to investigate the performance of one-vs-all-based and one-vs-one-based ODD features for multi-class classification. The statistical results show that the classification accuracy based on ODD features outperforms that of the state-of-the-art feature extraction methods.     

Efficient parameter learning of Bayesian network classifiers

Recent advances have demonstrated substantial benefits from learning with both generative and discriminative parameters. On the one hand, generative approaches address the estimation of the parameters of the joint distribution—\(\mathrm{P}(y,\mathbf{x})\), which for most network types is very computationally efficient (a notable exception to this are Markov networks) and on the other hand, discriminative approaches address the estimation of the parameters of the posterior distribution—and, are more effective for classification, since they fit \(\mathrm{P}(y|\mathbf{x})\) directly. However, discriminative approaches are less computationally efficient as the normalization factor in the conditional log-likelihood precludes the derivation of closed-form estimation of parameters. This paper introduces a new discriminative parameter learning method for Bayesian network classifiers that combines in an elegant fashion parameters learned using both generative and discriminative methods. The proposed method is discriminative in nature, but uses estimates of generative probabilities to speed-up the optimization process. A second contribution is to propose a simple framework to characterize the parameter learning task for Bayesian network classifiers. We conduct an extensive set of experiments on 72 standard datasets and demonstrate that our proposed discriminative parameterization provides an efficient alternative to other state-of-the-art parameterizations.     

Dynamic classifier selection for One-vs-One strategy: Avoiding non-competent classifiers

The One-vs-One strategy is one of the most commonly used decomposition technique to overcome multi-class classification problems; this way, multi-class problems are divided into easier-to-solve binary classification problems considering pairs of classes from the original problem, which are then learned by independent base classifiers.The way of performing the division produces the so-called non-competence. This problem occurs whenever an instance is classified, since it is submitted to all the base classifiers although the outputs of some of them are not meaningful (they were not trained using the instances from the class of the instance to be classified). This issue may lead to erroneous classifications, because in spite of their incompetence, all classifiers' decisions are usually considered in the aggregation phase.In this paper, we propose a dynamic classifier selection strategy for One-vs-One scheme that tries to avoid the non-competent classifiers when their output is probably not of interest. We consider the neighborhood of each instance to decide whether a classifier may be competent or not. In order to verify the validity of the proposed method, we will carry out a thorough experimental study considering different base classifiers and comparing our proposal with the best performer state-of-the-art aggregation within each base classifier from the five Machine Learning paradigms selected. The findings drawn from the empirical analysis are supported by the appropriate statistical analysis.     

Adaptive semi-supervised learning on labeled and unlabeled data with different distributions

Developing methods for designing good classifiers from labeled samples whose distribution is different from that of test samples is an important and challenging research issue in the fields of machine learning and its application. This paper focuses on designing semi-supervised classifiers with a high generalization ability by using unlabeled samples drawn by the same distribution as the test samples and presents a semi-supervised learning method based on a hybrid discriminative and generative model. Although JESS-CM is one of the most successful semi-supervised classifier design frameworks based on a hybrid approach, it has an overfitting problem in the task setting that we consider in this paper. We propose an objective function that utilizes both labeled and unlabeled samples for the discriminative training of hybrid classifiers and then expect the objective function to mitigate the overfitting problem. We show the effect of the objective function by theoretical analysis and empirical evaluation. Our experimental results for text classification using four typical benchmark test collections confirmed that with our task setting in most cases, the proposed method outperformed the JESS-CM framework. We also confirmed experimentally that the proposed method was useful for obtaining better performance when classifying data samples into either known or unknown classes, which were included in given labeled samples or not, respectively.     

Cost-Sensitive back-propagation neural networks with binarization techniques in addressing multi-class problems and non-competent classifiers

Multi-class classification problems can be addressed by using decomposition strategy. One of the most popular decomposition techniques is the One-vs-One (OVO) strategy, which consists of dividing multi-class classification problems into as many as possible pairs of easier-to-solve binary sub-problems. To discuss the presence of classes with different cost, in this paper, we examine the behavior of an ensemble of Cost-Sensitive Back-Propagation Neural Networks (CSBPNN) with OVO binarization techniques for multi-class problems. To implement this, the original multi-class cost-sensitive problem is decomposed into as many sub-problems as possible pairs of classes and each sub-problem is learnt in an independent manner using CSBPNN. Then a combination method is used to aggregate the binary cost-sensitive classifiers. To verify the synergy of the binarization technique and CSBPNN for multi-class cost-sensitive problems, we carry out a thorough experimental study. Specifically, we first develop the study to check the effectiveness of the OVO strategy for multi-class cost-sensitive learning problems. Then, we develop a comparison of several well-known aggregation strategies in our scenario. Finally, we explore whether further improvement can be achieved by using the management of non-competent classifiers. The experimental study is performed with three types of cost matrices and proper statistical analysis is employed to extract the meaningful findings.     

Constrained Minimization and Discriminative Training for Natural Language Call Routing

This paper presents a combination strategy of multiple individual routing classifiers to improve classification accuracy in natural language call routing applications. Since errors of individual classifiers in the ensemble should somehow be uncorrelated, we propose a combination strategy where the combined classifier accuracy is a function of the accuracy of individual classifiers and also the correlation between their classification errors. We show theoretically and empirically that our combination strategy, named the constrained minimization technique, has a good potential in improving the classification accuracy of single classifiers. We also show how discriminative training, more specifically the generalized probabilistic descent (GPD) algorithm, can be of benefit to further boost the performance of routing classifiers. The GPD algorithm has the potential to consider both positive and negative examples during training to minimize the classification error and increase the score separation of the correct from competing hypotheses. Some parameters become negative when using the GPD algorithm, resulting from suppressive learning not traditionally possible; important antifeatures are thus obtained. Experimental evaluation is carried on a banking call routing task and on switchboard databases with a set of 23 and 67 destinations, respectively. Results show either the GPD or constrained minimization technique outperform the accuracy of baseline classifiers by 44% when applied separately. When the constrained minimization technique is added on top of GPD, we show an additional 15% reduction in the classification error rate.     

Generative and discriminative learning by CL-Net.

This correspondence presents a two-stage classification learning algorithm. The first stage approximates the class-conditional distribution of a discrete space using a separate mixture model, and the second stage investigates the class posterior probabilities by training a network. The first stage explores the generative information that is inherent in each class by using the Chow-Liu (CL) method, which approximates high-dimensional probability with a tree structure, namely, a dependence tree, whereas the second stage concentrates on discriminative learning to distinguish between classes. The resulting learning algorithm integrates the advantages of both generative learning and discriminative learning. Because it uses CL dependence-tree estimation, we call our algorithm CL-Net. Empirical tests indicate that the proposed learning algorithm makes significant improvements when compared with the related classifiers that are constructed by either generative learning or discriminative learning.     

DRCW-OVO: Distance-based relative competence weighting combination for One-vs-One strategy in multi-class problems

One-vs-One strategy is a common and established technique in Machine Learning to deal with multi-class classification problems. It consists of dividing the original multi-class problem into easier-to-solve binary subproblems considering each possible pair of classes. Since several classifiers are learned, their combination becomes crucial in order to predict the class of new instances. Due to the division procedure a series of difficulties emerge at this stage, such as the non-competence problem. Each classifier is learned using only the instances of its corresponding pair of classes, and hence, it is not competent to classify instances belonging to the rest of the classes; nevertheless, at classification time all the outputs of the classifiers are taken into account because the competence cannot be known a priori (the classification problem would be solved). On this account, we develop a distance-based combination strategy, which weights the competence of the outputs of the base classifiers depending on the closeness of the query instance to each one of the classes. Our aim is to reduce the effect of the non-competent classifiers, enhancing the results obtained by the state-of-the-art combinations for One-vs-One strategy. We carry out a thorough experimental study, supported by the proper statistical analysis, showing that the results obtained by the proposed method outperform, both in terms of accuracy and kappa measures, the previous combinations for One-vs-One strategy.     

基于概率投票策略的多类支持向量机及应用

传统的支持向量机是基于两类问题提出的,如何将其有效地推广至多类分类仍是一个研究的热点问题。在分析比较现有支持向量机多类分类OVO方法存在的问题及缺点的基础上,该文提出一种新的基于概率投票策略的多类分类方法。在该策略中,充分考虑了OVO方法中各个两类支持向量机分类器的差异,并将该差异反映到投票分值上。所提多类支持向量机方法不仅具有较好的分类性能,而且有效解决了传统投票策略中存在的拒分区域问题。将基于概率投票的多分类支持向量机作为关键技术应用于实际齿轮箱故障诊断,并与传统投票策略的结果进行对比,表明所提方法的上述优点。