Evolutionary constructive induction

Feature construction in classification is a preprocessing step in which one or more new attributes are constructed from the original attribute set, the object being to construct features that are more predictive than the original feature set. Genetic programming allows the construction of nonlinear combinations of the original features. We present a comprehensive analysis of genetic programming (GP) used for feature construction, in which four different fitness functions are used by the GP and four different classification techniques are subsequently used to build the classifier. Comparisons are made of the error rates and the size and complexity of the resulting trees. We also compare the overall performance of GP in feature construction with that of GP used directly to evolve a decision tree classifier, with the former proving to be a more effective use of the evolutionary paradigm.     

Unsupervised feature selection with graph learning via low-rank constraint

Feature selection is one of the most important machine learning procedure, and it has been successfully applied to make a preprocessing before using classification and clustering methods. High-dimensional features often appear in big data, and it’s characters block data processing. So spectral feature selection algorithms have been increasing attention by researchers. However, most feature selection methods, they consider these tasks as two steps, learn similarity matrix from original feature space (may be include redundancy for all features), and then conduct data clustering. Due to these limitations, they do not get good performance on classification and clustering tasks in big data processing applications. To address this problem, we propose an Unsupervised Feature Selection method with graph learning framework, which can reduce the redundancy features influence and utilize a low-rank constraint on the weight matrix simultaneously. More importantly, we design a new objective function to handle this problem. We evaluate our approach by six benchmark datasets. And all empirical classification results show that our new approach outperforms state-of-the-art feature selection approaches.     

An adaptive memetic algorithm for feature selection using proximity graphs

We propose a multivariate feature selection method that uses proximity graphs for assessing the quality of feature subsets. Initially, a complete graph is built, where nodes are the samples, and edge weights are calculated considering only the selected features. Next, a proximity graph is constructed on the basis of these weights and different fitness functions, calculated over the proximity graph, to evaluate the quality of the selected feature set. We propose an iterative methodology on the basis of a memetic algorithm for exploring the space of possible feature subsets aimed at maximizing a quality score. We designed multiple local search strategies, and we used an adaptive strategy for automatic balancing between the global and local search components of the memetic algorithm. The computational experiments were carried out using four well‐known data sets. We investigate the suitability of three different proximity graphs (minimum spanning tree, k‐nearest neighbors, and relative neighborhood graph) for the proposed approach. The selected features have been evaluated using a total of 49 classification methods from an open‐source data mining and machine learning package (WEKA). The computational results show that the proposed adaptive memetic algorithm can perform better than traditional genetic algorithms in finding more useful feature sets. Finally, we establish the competitiveness of our approach by comparing it with other well‐known feature selection methods.     

基于遗传算法的蛋白质质谱数据特征选择

针对蛋白质质谱数据在降维、分类及生物标记物识别过程中存在的问题,提出一种基于遗传算法的特征选择方法,介绍几种常用的相关策略,包括基于排列和精英保留的随机通用采样选择策略和基于自适应变肄率的均匀变异策略,给出2个适应度函数——封装器函数与多变元筛选器函数,将它们引入遗传算法中,并进行性能测试与比较。实验结果表明,基于封装器的遗传算法性能优于其他特征选择算法,而基于多变元筛选器的遗传算法性能优于单变元筛选器算法。     

Building lightweight intrusion detection system using wrapper-based feature selection mechanisms

Intrusion Detection System (IDS) is an important and necessary component in ensuring network security and protecting network resources and network infrastructures. How to build a lightweight IDS is a hot topic in network security. Moreover, feature selection is a classic research topic in data mining and it has attracted much interest from researchers in many fields such as network security, pattern recognition and data mining. In this paper, we effectively introduced feature selection methods to intrusion detection domain. We propose a wrapper-based feature selection algorithm aiming at building lightweight intrusion detection system by using modified random mutation hill climbing (RMHC) as search strategy to specify a candidate subset for evaluation, as well as using modified linear Support Vector Machines (SVMs) iterative procedure as wrapper approach to obtain the optimum feature subset. We verify the effectiveness and the feasibility of our feature selection algorithm by several experiments on KDD Cup 1999 intrusion detection dataset. The experimental results strongly show that our approach is not only able to speed up the process of selecting important features but also to yield high detection rates. Furthermore, our experimental results indicate that intrusion detection system with feature selection algorithm has better performance than that without feature selection algorithm both in detection performance and computational cost.     

基于密度估计的逻辑回归模型

介绍了一种基于密度的逻辑回归（Density-based logistic regression,DLR）分类模型以解决逻辑回归中非线性分类的问题. 其主要思想是根据Nadarays-Watson密度估计将训练数据映射到特定的特征空间,然后组建优化模型优化特征权重以及Nadarays-Watson 密度估计算法的宽度. 其主要优点在于:它不仅优于标准的逻辑回归,而且优于基于径向基函数（Radial basis function,RBF）内核的核逻辑回归（Kernel logistic regression,KLR）. 特别是与核逻辑回归分析和支持向量机（Support vector machine,SVM）相比,该方法不仅达到更好的分类精度,而且有更好的时间效率. 该方法的另一个显著优点是,它可以很自然地扩展到数值类型和分类型混合的数据集中. 除此之外,该方法和逻辑回归（Logistic regression,LR）一样,有同样的模型可解释的优点,这恰恰是其他如核逻辑回归分析和支持向量机所不具备的.     

Exhaustive and heuristic search approaches for learning a software defect prediction model

In this paper, we propose a software defect prediction model learning problem (SDPMLP) where a classification model selects appropriate relevant inputs, from a set of all available inputs, and learns the classification function. We show that the SDPMLP is a combinatorial optimization problem with factorial complexity, and propose two hybrid exhaustive search and probabilistic neural network (PNN), and simulated annealing (SA) and PNN procedures to solve it. For small size SDPMLP, exhaustive search PNN works well and provides an (all) optimal solution(s). However, for large size SDPMLP, the use of exhaustive search PNN approach is not pragmatic and only the SA–PNN allows us to solve the SDPMLP in a practical time limit. We compare the performance of our hybrid approaches with traditional classification algorithms and find that our hybrid approaches perform better than traditional classification algorithms.     

Comparison study on the performance of the multi classifiers with hybrid optimal features selection method for medical data diagnosis

Features subset selection (FSS) generally plays an essential role in the implementation of data mining, particularly in the field of high-dimensional medical data analysis, as well as in supplying early detection with essential features and high accuracy. The latest modern feature selection models are now using the ability of optimization algorithms for extracting features of particular properties to get the highest accuracy performance possible. Many of the optimization algorithms, such as genetic algorithm, often use the required parameters that would need to be adjusted for better results. For the function selection procedure, tuning these parameter values is a difficult challenge. In this paper, a new wrapper-based feature selection approach called binary teaching learning based optimization (BTLBO) is introduced. The binary teaching learning based optimization (BTLBO) is among the most sophisticated meta-heuristic method which does not involve any specific algorithm parameters. It requires only standard process parameters such as population size and a number of iterations to extract a set of features selected from a data. This is a demanding process, to achieve the best possible set of features would be to use a method which is independent of the method controlling parameters. This paper introduces a new modified binary teaching–learning-based optimization (NMBTLBO) as a technique to select subset features and demonstrate support vector machine (SVM) accuracy of binary identification as a fitness function for the implementation of the feature subset selection process. The new proposed algorithm NMBTLBO contains two steps: first, the new updating procedure, second, the new method to select the primary teacher in teacher phase in binary teaching-learning based on optimization algorithm. The proposed technique NMBTLBO was used to classify the rheumatic disease datasets collected from Baghdad Teaching Hospital Outpatient Rheumatology Clinic during 2016–2018. Compared with the original BTLBO algorithm, the improved NMBTLBO algorithm has achieved a major difference in accuracy. Validation was carried out by testing the accuracy of four classification methods: K-nearest neighbors, decision trees, support vector machines and K-means. Study results showed that the classification accuracy of the four methods was increased for the proposed method of selection of features (NMBTLBO) compared to the BTLBO algorithm. SVM classifier provided 89% accuracy of BTLBO-SVM and 95% with NMBTLBO –SVM. Decision trees set the values of 94% with BTLBO-SVM and 95% with the feature selection of NMBTLBO-SVM. The analysis indicates that the latest method (NMBTLBO) enhances classification accuracy.

     

Automatic detection of speaker state: Lexical,prosodic, and phonetic approaches to level-of-interest and intoxication classification

Traditional studies of speaker state focus primarily upon one-stage classification techniques using standard acoustic features. In this article, we investigate multiple novel features and approaches to two recent tasks in speaker state detection: level-of-interest (LOI) detection and intoxication detection. In the task of LOI prediction, we propose a novel Discriminative TFIDF feature to capture important lexical information and a novel Prosodic Event detection approach using AuToBI; we combine these with acoustic features for this task using a new multilevel multistream prediction feedback and similarity-based hierarchical fusion learning approach. Our experimental results outperform published results of all systems in the 2010 Interspeech Paralinguistic Challenge – Affect Subchallenge. In the intoxication detection task, we evaluate the performance of Prosodic Event-based, phone duration-based, phonotactic, and phonetic-spectral based approaches, finding that a combination of the phonotactic and phonetic-spectral approaches achieve significant improvement over the 2011 Interspeech Speaker State Challenge – Intoxication Subchallenge baseline. We discuss our results using these new features and approaches and their implications for future research.     

Recognition of human actions using CNN-GWO: a novel modeling of CNN for enhancement of classification performance

Recognizing human actions from unconstrained videos turns to be a major challenging task in computer visualization approaches due to decreased accuracy in the feature classification performance. Therefore to improve the classification performance it is essential to minimize the ‘classification’ errors. Here, in this work, we propose a hybrid CNN-GWO approach for the recognition of human actions from the unconstrained videos. The weight initializations for the proposed deep Convolutional Neural Network (CNN) classifiers highly depend on the generated solutions of GWO (Grey Wolf Optimization) algorithm, which in turn minimizes the ‘classification’ errors. The action bank and local spatio-temporal features are generated for a video and fed into the ‘CNN’ classifiers. The ‘CNN’ classifiers are trained by a gradient descent algorithm to detect a ‘local minimum’ during the fitness computation of GWO ‘search agents’. The GWO algorithms ‘global search’ capability as well as the gradient descent algorithms ‘local search’ capabilities are subjected for the identification of a solution which is nearer to the global optimum. Finally, the classification performance can be further enhanced by fusing the classifiers evidences produced by the GWO algorithm. The proposed classification frameworks efficiency for the recognition of human actions is evaluated with the help of four achievable action recognition datasets namely HMDB51, UCF50, Olympic Sports and Virat Release 2.0. The experimental validation of our proposed approach shows better achievable results on the recognition of human actions with 99.9% recognition accuracy.     

面向软件缺陷个数预测的混合式特征选择方法*

针对软件缺陷数据集中不相关特征和冗余特征会降低软件缺陷个数预测模型的性能的问题,提出了一种面向软件缺陷个数预测的混合式特征选择方法-HFSNFP。首先,利用ReliefF算法计算每个特征与缺陷个数之间的相关性,选出相关性最高的m个特征;然后,基于特征之间的关联性利用谱聚类对这m个特征进行聚类;最后,利用基于包裹式特征选择思想从每个簇中依次挑选最相关的特征形成最终的特征子集。实验结果表明,相比于已有的五种过滤式特征选择方法,HFSNFP方法在提高预测率的同时降低了误报率,且G-measure与RMSE度量值更佳;相比于已有的两种包裹式特征选择方法,HFSNFP方法在保证了缺陷个数预测性能的同时可以显著降低特征选择的时间。     

基于语义感知的图像美学质量评估方法

当前图像美学质量评估的研究主要基于图像的视觉内容来给出评价结果,忽视了美感是人的认知活动的事实,在评价时没有考虑用户对图像语义信息的理解。为了解决这一问题,提出了一种基于语义感知的图像美学质量评估方法,将图像的物体类别信息以及场景类别信息也用于图像美学质量评估。运用迁移学习的思想,构建了一种可以融合图像多种特征的混合网络。对于每一幅输入图像,该网络可以分别提取出其物体类别特征、场景类别特征以及美学特征,并将这三种特征进行高质量的融合,以达到更好的图像美学质量评估效果。该方法在AVA数据集上的分类准确率达到89.5%,相对于传统方法平均提高了19.9%,在CUHKPQ数据集上的泛化性能也有了很大提升。实验结果表明,所提方法在图像美学质量评估问题上,能够取得更好的分类性能。     

Selection of Relevant Visual Feature Sets for Enhanced Depression Detection using Incremental Linear Discriminant Analysis

Presently, while automated depression diagnosis has made great progress, most of the recent works have focused on combining multiple modalities rather than strengthening a single one. In this research work, we present a unimodal framework for depression detection based on facial expressions and facial motion analysis. We investigate a wide set of visual features extracted from different facial regions. Due to high dimensionality of the obtained feature sets, identification of informative and discriminative features is a challenge. This paper suggests a hybrid dimensionality reduction approach which leverages the advantages of the filter and wrapper methods. First, we use a univariate filter method, Fisher Discriminant Ratio, to initially reduce the size of each feature set. Subsequently, we propose an Incremental Linear Discriminant Analysis (ILDA) approach to find an optimal combination of complementary and relevant feature sets. We compare the performance of the proposed ILDA with the batch-mode LDA and also the Composite Kernel based Support Vector Machine (CKSVM) method. The experiments conducted on the Distress Analysis Interview Corpus Wizard-of-Oz (DAIC-WOZ) dataset demonstrate that the best depression classification performance is obtained by using different feature extraction methods in combination rather than individually. ILDA generates better depression classification results in comparison to the CKSVM. Moreover, ILDA based wrapper feature selection incurs lower computational cost in comparison to the CKSVM and the batch-mode LDA methods. The proposed framework significantly improves the depression classification performance, with an F1 Score of 0.805, which is better than all the video based depression detection models suggested in literature, for the DAIC-WOZ dataset. Salient facial regions and well performing visual feature extraction methods are also identified.

     

Developing new fitness functions in genetic programming for classification with unbalanced data

Machine learning algorithms such as genetic programming (GP) can evolve biased classifiers when data sets are unbalanced. Data sets are unbalanced when at least one class is represented by only a small number of training examples (called the minority class) while other classes make up the majority. In this scenario, classifiers can have good accuracy on the majority class but very poor accuracy on the minority class(es) due to the influence that the larger majority class has on traditional training criteria in the fitness function. This paper aims to both highlight the limitations of the current GP approaches in this area and develop several new fitness functions for binary classification with unbalanced data. Using a range of real-world classification problems with class imbalance, we empirically show that these new fitness functions evolve classifiers with good performance on both the minority and majority classes. Our approaches use the original unbalanced training data in the GP learning process, without the need to artificially balance the training examples from the two classes (e.g., via sampling).     

Domain-independent feature extraction for multi-classification using multi-objective genetic programming

We propose three model-free feature extraction approaches for solving the multiple class classification problem; we use multi-objective genetic programming (MOGP) to derive (near-)optimal feature extraction stages as a precursor to classification with a simple and fast-to-train classifier. Statistically-founded comparisons are made between our three proposed approaches and seven conventional classifiers over seven datasets from the UCI Machine Learning database. We also make comparisons with other reported evolutionary computation techniques. On almost all the benchmark datasets, the MOGP approaches give better or identical performance to the best of the conventional methods. Of our proposed MOGP-based algorithms, we conclude that hierarchical feature extraction performs best on multi-classification problems.     

Feature generation using genetic programming with comparative partner selection for diabetes classification

The ultimate aim of this research is to facilitate the diagnosis of diabetes, a rapidly increasing disease in the world. In this research a genetic programming (GP) based method has been used for diabetes classification. GP has been used to generate new features by making combinations of the existing diabetes features, without prior knowledge of the probability distribution. The proposed method has three stages: features selection is performed at the first stage using t-test, Kolmogorov–Smirnov test, Kullback–Leibler divergence test, F-score selection, and GP. The results of feature selection methods are used to prepare an ordered list of original features where features are arranged in decreasing order of importance. Different subsets of original features are prepared by adding features one by one in each subset using sequential forward selection method according to the ordered list. At the second stage, GP is used to generate new features from each subset of original diabetes features, by making non-linear combinations of the original features. A variation of GP called GP with comparative partner selection (GP-CPS), utilising the strengths and the weaknesses of GP generated features, has been used at the second stage. The performance of GP generated features for classification is tested using the k-nearest neighbor and support vector machine classifiers at the last stage. The results and their comparisons with other methods demonstrate that the proposed method exhibits superior performance over other recent methods.     

Robust text detection in natural scenes using text geometry and visual appearance

This paper proposes a new two-phase approach to robust text detection by integrating the visual appearance and the geometric reasoning rules. In the first phase, geometric rules are used to achieve a higher recall rate. Specifically, a robust stroke width transform (RSWT) feature is proposed to better recover the stroke width by additionally considering the cross of two strokes and the continuousness of the letter border. In the second phase, a classification scheme based on visual appearance features is used to reject the false alarms while keeping the recall rate. To learn a better classifier from multiple visual appearance features, a novel classification method called double soft multiple kernel learning (DS-MKL) is proposed. DS-MKL is motivated by a novel kernel margin perspective for multiple kernel learning and can effectively suppress the influence of noisy base kernels. Comprehensive experiments on the benchmark ICDAR2005 competition dataset demonstrate the effectiveness of the proposed two-phase text detection approach over the state-of-the-art approaches by a performance gain up to 4.4% in terms of F-measure.     

Online Feature Selection of Class Imbalance via PA Algorithm

Imbalance classification techniques have been frequently applied in many machine learning application domains where the number of the majority (or positive) class of a dataset is much larger than that of the minority (or negative) class. Meanwhile, feature selection (FS) is one of the key techniques for the high-dimensional classification task in a manner which greatly improves the classification performance and the computational efficiency. However, most studies of feature selection and imbalance classification are restricted to off-line batch learning, which is not well adapted to some practical scenarios. In this paper, we aim to solve high-dimensional imbalanced classification problem accurately and efficiently with only a small number of active features in an online fashion, and we propose two novel online learning algorithms for this purpose. In our approach, a classifier which involves only a small and fixed number of features is constructed to classify a sequence of imbalanced data received in an online manner. We formulate the construction of such online learner into an optimization problem and use an iterative approach to solve the problem based on the passive-aggressive (PA) algorithm as well as a truncated gradient (TG) method. We evaluate the performance of the proposed algorithms based on several real-world datasets, and our experimental results have demonstrated the effectiveness of the proposed algorithms in comparison with the baselines.     

Particle swarm optimization-based feature selection in sentiment classification

Sentiment classification is one of the important tasks in text mining, which is to classify documents according to their opinion or sentiment. Documents in sentiment classification can be represented in the form of feature vectors, which are employed by machine learning algorithms to perform classification. For the feature vectors, the feature selection process is necessary. In this paper, we will propose a feature selection method called fitness proportionate selection binary particle swarm optimization (F-BPSO). Binary particle swarm optimization (BPSO) is the binary version of particle swam optimization and can be applied to feature selection domain. F-BPSO is a modification of BPSO and can overcome the problems of traditional BPSO including unreasonable update formula of velocity and lack of evaluation on every single feature. Then, some detailed changes are made on the original F-BPSO including using fitness sum instead of average fitness in the fitness proportionate selection step. The modified method is, thus, called fitness sum proportionate selection binary particle swarm optimization (FS-BPSO). Moreover, further modifications are made on the FS-BPSO method to make it more suitable for sentiment classification-oriented feature selection domain. The modified method is named as SCO-FS-BPSO where SCO stands for “sentiment classification-oriented”. Experimental results show that in benchmark datasets original F-BPSO is superior to traditional BPSO in feature selection performance and FS-BPSO outperforms original F-BPSO. Besides, in sentiment classification domain, SCO-FS-BPSO which is modified specially for sentiment classification is superior to traditional feature selection methods on subjective consumer review datasets.     

A feature construction method for general object recognition

This paper presents a novel approach for object detection using a feature construction method called Evolution-COnstructed (ECO) features. Most other object recognition approaches rely on human experts to construct features. ECO features are automatically constructed by uniquely employing a standard genetic algorithm to discover series of transforms that are highly discriminative. Using ECO features provides several advantages over other object detection algorithms including: no need for a human expert to build feature sets or tune their parameters, ability to generate specialized feature sets for different objects, and no limitations to certain types of image sources. We show in our experiments that ECO features perform better or comparable with hand-crafted state-of-the-art object recognition algorithms. An analysis is given of ECO features which includes a visualization of ECO features and improvements made to the algorithm.