Standard UCI Datasets Research Articles

Research on improved text classification method based on combined weighted model

SummaryText classification is very important in information retrieval, but the traditional text classification model has many problems, such as the feature dimension disaster, the lack of semantic features, etc. Aiming at the problems, this paper proposes an improved TFIDF model combined with the Word2vec model for weighing word vectors. In view of the inability of the Word2vec model to distinguish the importance of words with the text, TFIDF is further introduced to weighing Word2vec word vectors to achieve a weighted Word2vec classification model. For data preprocessing, we optimized the traditional StringToWordVector algorithm. The main improvement of StringToWordVector is the introduction to a new algorithm of stem extraction. First, this paper gives a simple description of the basic steps and algorithms of traditional text classification, and then, the ideas and steps of the improved StringToWordVector algorithm are proposed. Finally, experimental results using our improved algorithm are tested for four different data sets (WEBO_SINA and three standard UCI data sets). The experimental results show that the improved StringToWordVector algorithm combined with the combined weighted model has higher classification accuracy, recall, and F1 values than the traditional text classification model only using the Word2vec model or using TFIDF. The experimental results are satisfactory.

Data driven particle size estimation of hematite grinding process using stochastic configuration network with robust technique

As a production quality index of hematite grinding process, particle size (PS) is hard to be measured in real time. To achieve the PS estimation, this paper proposes a novel data driven model of PS using stochastic configuration network (SCN) with robust technique, namely, robust SCN (RSCN). Firstly, this paper proves the universal approximation property of RSCN with weighted least squares technique. Secondly, three robust algorithms are presented by employing M-estimation with Huber loss function, M-estimation with interquartile range (IQR) and nonparametric kernel density estimation (NKDE) function respectively to set the penalty weight. Comparison experiments are first carried out based on the UCI standard data sets to verify the effectiveness of these methods, and then the data-driven PS model based on the robust algorithms are established and verified. Experimental results show that the RSCN has an excellent performance for the PS estimation.

Binary grasshopper optimisation algorithm approaches for feature selection problems

Feature Selection (FS) is a challenging machine learning-related task that aims at reducing the number of features by removing irrelevant, redundant and noisy data while maintaining an acceptable level of classification accuracy. FS can be considered as an optimisation problem. Due to the difficulty of this problem and having a large number of local solutions, stochastic optimisation algorithms are promising techniques to solve this problem. As a seminal attempt, binary variants of the recent Grasshopper Optimisation Algorithm (GOA) are proposed in this work and employed to select the optimal feature subset for classification purposes within a wrapper-based framework. Two mechanisms are employed to design a binary GOA, the first one is based on Sigmoid and V-shaped transfer functions, and will be indicated by BGOA-S and BGOA-V, respectively. While the second mechanism uses a novel technique that combines the best solution obtained so far. In addition, a mutation operator is employed to enhance the exploration phase in BGOA algorithm (BGOA-M). The proposed methods are evaluated using 25 standard UCI datasets and compared with 8 well-regarded metaheuristic wrapper-based approaches, and six well known filter-based (e.g., correlation FS) approaches. The comparative results show the superior performance of the BGOA and BGOA-M methods compared to other similar techniques in the literature.

A Roughset Based Ensemble Framework for Network Intrusion Detection System

Designing an effective network intrusion detection system is becoming an increasingly difficult task as the sophistication of the attacks have been increasing every day. Usage of machine learning approaches has been proving beneficial in such situations. Models may be developed based on patterns differentiating attack traffic from network traffic to gain insight into the network activity to identify and report attacks. In this article, an ensemble framework based on roughsets is used to efficiently identify attacks in a multi-class scenario. The proposed methodology is validated on benchmark KDD Cup '99 and NSL_KDD network intrusion detection datasets as well as six other standard UCI datasets. The experimental results show that proposed technique RST achieved better detection rate with low false alarm rate compared to bagging and RSM.

Developed Density Peak Clustering With Support Vector Data Description for Access Network Intrusion Detection

Access network intrusion detection (ANID) is an indispensable role in guaranteeing information security, in which data-classification is a key operating procedure. In this paper, we propose an algorithm of the developed density peak clustering with support vector data description (SVDD), and it can break through the bottleneck of processing high-dimensional data with non-uniform density in the ANID system. The cutoff distance $(dc)$ depending on experience in density peak clustering is then optimized by our defined coefficient called adjusted silhouette coefficient (ASIL). This can overcome the unfitness of traditional SIL due to multi-noise samples and clustering instability. In order to generate an accurate classifier, an improved particle swarm optimization algorithm is also developed to the parameters of SVDD. Finally, the proposed algorithm is evaluated through UCI standard data sets and KDD 99 data sets, and these results show higher accuracy and robustness compared with the conventional methods.

An Affinity Propagation Clustering Method Using Hybrid Kernel Function With LLE

Cluster analysis is important in data mining and clustering algorithms and has gained much attention during the last decade. However, it is a challenge to extract significant features from high-dimensional data and to rapidly provide satisfactory clustering results. This paper presents a new affinity propagation (AP) clustering method based on a hybrid kernel function with locally linear embedding, called LLE-HKAP, for the classification of gene expression datasets and standard UCI datasets. First, the locally linear embedding algorithm is used to reduce the dimension of the original dataset. Then, a novel AP clustering method based on a similarity measure with the hybrid kernel function is proposed. In this method, a new global kernel is defined that has high generalization ability. Meanwhile, a hybrid kernel function that linearly combines the proposed global kernel and the Gaussian kernel is defined to further enhance the learning ability of the global kernel. Moreover, the novel hybrid kernel is introduced to define a similarity measure and construct a similarity matrix of the AP clustering. Finally, the improved AP clustering algorithm is implemented on eight public gene expression datasets and eight standard UCI datasets for comparison with other related algorithms. The experimental results validate that our proposed clustering algorithm is efficient in terms of clustering accuracy and outperforms the currently available approaches with which it is compared.

An Adaptive Multiobjective Genetic Algorithm with Fuzzy c-Means for Automatic Data Clustering

This paper presents a fuzzy clustering method based on multiobjective genetic algorithm. The ADNSGA2-FCM algorithm was developed to solve the clustering problem by combining the fuzzy clustering algorithm (FCM) with the multiobjective genetic algorithm (NSGA-II) and introducing an adaptive mechanism. The algorithm does not need to give the number of clusters in advance. After the number of initial clusters and the center coordinates are given randomly, the optimal solution set is found by the multiobjective evolutionary algorithm. After determining the optimal number of clusters by majority vote method, the Jm value is continuously optimized through the combination of Canonical Genetic Algorithm and FCM, and finally the best clustering result is obtained. By using standard UCI dataset verification and comparing with existing single-objective and multiobjective clustering algorithms, the effectiveness of this method is proved.

A Novel Least Square Twin Support Vector Regression

This paper proposes a new method for regression named lp norm least square twin support vector regression (PLSTSVR), which is formulated by the idea of twin support vector regression (TSVR). Different from TSVR, our new model is an adaptive learning procedure with p-norm SVM ( $${{0<p\le 2}}$$ ), where p is viewed as an adjustable parameter and can be automatically chosen by data. An iterative algorithm is suggested to solve PLSTSVR efficiently. In each iteration, only a series systems of linear equations (LEs) are solved. Experiments carried out on several standard UCI datasets and synthetic datasets show the feasibility and effectiveness of the proposed method.

Handling binary classification problems with a priority class by using Support Vector Machines

A post-processing technique for Support Vector Machine (SVM) algorithms for binary classification problems is introduced in order to obtain adequate accuracy on a priority class (labelled as a positive class). That is, the true positive rate (or recall or sensitivity) is prioritized over the accuracy of the overall classifier. Hence, false negative (or Type I) errors receive greater consideration than false positive (Type II) errors during the construction of the model.This post-processing technique tunes the initial bias term once a solution vector is learned by using standard SVM algorithms in two steps: First, a fixed threshold is given as a lower bound for the recall measure; second, the true negative rate (or specificity) is maximized.Experiments, carried out on eleven standard UCI datasets, show that the modified SVM satisfies the aims for which it has been designed. Furthermore, results are comparable or better than those obtained when other state-of-the-art SVM algorithms and other usual metrics are considered.

Improved linear local tangent space alignment and its application to pattern recognition

Considering the drawbacks of the linear local tangent space alignment, a Semi-Supervised Neighbourhood Adaptive Linear Local Tangent Space Alignment (SSNA-LLTSA) is proposed. The distance metric combining the Cosine similarity and the Euclidean distance is used in the algorithm instead of the Euclidean distance, and the algorithm realises the semi-supervised learning and neighbourhood adaptive adjustment by integrating some of the known category information and the method of Parzen window density estimation into the dimensionality reduction process. The simulation experiment of UCI standard data sets and the pattern recognition example of hydraulic pump show that the redefined distance metric has better performance than the Euclidean distance, and SSNA-LLTSA can overcome the defect that LLTSA is unsupervised. Meanwhile, the capability of neighbourhood adaptive adjustment makes the algorithm find the low-dimensional manifold of the data sets more effectively, which can further improve the accuracy of pattern recognition.

A new cognitive filtering approach based on Freeman K3 Neural Networks

Huge volume of data over several domains demands the development of new more efficient tools for search, analysis, and interpretation. Clustering approaches represent an important step in exploring the internal structure and relationships in datasets. In this study, the cognitively motivated neural network Freeman K 3-set was applied as a filter to preprocess the data, achieving a better clustering performance. We combine K 3 with a variety of clustering algorithms commonly used, and tested its performance using standard UCI datasets and also datasets from social networks. A comprehensive evaluation using a number of cluster validation measures shows significant improvement in the overall performance of the K 3-based clustering method for social data sets, for two types of clustering validation measures. Additionally, K 3 filtering results in transparent representation of data, which leads to improved efficiency of data processing algorithms used.

Research on High-Dimensional Data Reduction

In this paper, features of high-dimensional data are analyzed, and existing problems of the Canonical Correlation Analysis (CCA) are analyzed for a single view of a full supervised view data. In order to improve CCA, we introduce the method of classifier and present a Classifying to Reduce Correlation Dimensionality (CRCD). Meanwhile, combining big interval learning method, we propose the big correlation analysis (BCA). At last, experiments are respectively conducted by using artificial data set and UCI standard data set. The result shows that methods are feasible and effective.

Feature Selection Using Relative Fuzzy Entropy and Ant Colony Optimization Applied to Real-time Intrusion Detection System

Intrusion Detection System (IDS) is one of the most important component of network defense mechanism. In an attempt to detect network attacks, network traffic features need to be identified and both attack and normal data need to be profiled. This paper proposes a set of network traffic features that can be extracted for Real-Time Intrusion Detection. This paper also proposes Fuzzy Entropy based heuristic for Ant Colony Optimization (ACO) in-order to search for global best smallest set of network traffic features for Real-Time Intrusion Detection Data set. The proposed feature reduction algorithm was tested on standard bench-mark UCI data sets, and found to be efficient. Further the algorithm was applied to Real-Time IDS data set and found to produce promising results.

Support Vector Machine Optimization Based On Magnetic Bacteria Optimization Algorithm

Classification performance of support vector machine (SVM) will be influenced by its model parameters. For this problem, a new method named magnetic bacteria optimization algorithm (MBOA) that optimizes the parameters of SVM is proposed. It is tested by the UCI standard data sets and compared with the other optimization algorithms, such as particle swarm optimization (PSO). Experimental results show that the MBOA can optimize the parameters of SVM well and has better performance than the compared algorithms.

Finding rough set reducts with fish swarm algorithm

Rough set theory is one of the effective methods to feature selection which can preserve the characteristics of the original features by deleting redundant information. The main idea of rough set approach to feature selection is to find a globally minimal reduct, the smallest set of features keeping important information of the original set of features. Rough set theory has been used as a dataset preprocessor with much success, but current approaches to feature selection are inadequate for finding a globally minimal reduct. In this paper, we propose a novel rough set based method to feature selection using fish swarm algorithm. The fish swarm algorithm is a new intelligent swarm modeling approach that consists primarily of searching, swarming, and following behaviors. It is attractive for feature selection since fish swarms can discover the best combination of features as they swim within the subset space. In our proposed algorithm, a minimal subset can be located and verified. To show the efficiency of our algorithm, we carry out numerical experiments based on some standard UCI datasets. The results demonstrate that our algorithm can provide an efficient tool for finding a minimal subset of the features without information loss.

Feature-Weighted Local Support Vector Machine of Particle Swarm Optimization

Local support vector machine gives the feature same weight in classification. In fact, many datasets have some weak or irrelevant features related to the classification. Thus giving features same weight may reduce the classification accuracy of local support vector machine.This paper puts forward a new local support vector machine that the feature weight is optimized by PSO (Particle Swarm Optimization), it is tested on the international standard UCI data sets and the images of tree taxonomy data sets, the results show that the accuracy of the algorithm we proposed is better than the general local support vector machine.

Sparse least square twin support vector machine with adaptive norm

By promoting the parallel hyperplanes to non-parallel ones in SVM, twin support vector machines (TWSVM) have attracted more attention. There are many modifications of them. However, most of the modifications minimize the loss function subject to the I 2-norm or I 1-norm penalty. These methods are non-adaptive since their penalty forms are fixed and pre-determined for any types of data. To overcome the above shortcoming, we propose l p norm least square twin support vector machine (l p LSTSVM). Our new model is an adaptive learning procedure with l p -norm (0<p<1), where p is viewed as an adjustable parameter and can be automatically chosen by data. By adjusting the parameter p, l p LSTSVM can not only select relevant features but also improve the classification accuracy. The solutions of the optimization problems in l p LSTSVM are obtained by solving a series systems of linear equations (LEs) and the lower bounds of the solution is established which is extremely helpful for feature selection. Experiments carried out on several standard UCI data sets and synthetic data sets show the feasibility and effectiveness of the proposed method.

An entropy-based uncertainty measurement approach in neighborhood systems

Uncertainty measures can provide us with principled methodologies to analyze uncertain data and unveil the substantive characteristics of the data sets. Accuracy and roughness proposed by Pawlak are two main tools to deal with uncertainty measurement issue in rough set theory. Many uncertainty measure methodologies for discrete-valued information systems or discrete-valued decision systems have been developed. However, there are only limited on the uncertainty measurement for neighborhood systems. In this paper, we address the issues of uncertainty of a neighborhood system and extend the traditional accuracy and roughness measures to deal with neighborhood systems. In particular, a concept called neighborhood entropy is first introduced to evaluate the uncertainty of a neighborhood information system. Consequently, the entropy-based roughness and approximation roughness measures of neighborhood system are presented. Theoretical analysis indicates that entropy-based measures can be used to evaluate the uncertainty in neighborhood systems. Experiments are conducted on artificial data sets and standard UCI data sets to test our proposed methodologies. Results demonstrate that the entropy-based measures are effective and valid for evaluating the uncertainty of neighborhood systems.

An exact feature selection algorithm based on rough set theory

Feature reduction based on rough set theory is an effective feature selection method in pattern recognition applications. Finding a minimal subset of the original features is inherent in rough set approach to feature selection. As feature reduction is a Nondeterministic Polynomial‐time‐hard problem, it is necessary to develop fast optimal or near‐optimal feature selection algorithms. This article aims to propose an exact feature selection algorithm in rough set that is efficient in terms of computation time. The proposed algorithm begins the examination of a solution tree by a breadth‐first strategy. The pruned nodes are held in a version of the trie data structure. Based on the monotonic property of dependency degree, all subsets of the pruned nodes cannot be optimal solutions. Thus, by detecting these subsets in trie, it is not necessary to calculate their dependency degree. The search on the tree continues until the optimal solution is found. This algorithm is improved by selecting an initial search level determined by the hill‐climbing method instead of searching the tree from the level below the root. The length of the minimal reduct and the size of data set can influence which starting search level is more efficient. The experimental results using some of the standard UCI data sets, demonstrate that the proposed algorithm is effective and efficient for data sets with more than 30 features. © 2014 Wiley Periodicals, Inc. Complexity 20: 50–62, 2015

A New Optimized GA-RBF Neural Network Algorithm

When confronting the complex problems, radial basis function (RBF) neural network has the advantages of adaptive and self-learning ability, but it is difficult to determine the number of hidden layer neurons, and the weights learning ability from hidden layer to the output layer is low; these deficiencies easily lead to decreasing learning ability and recognition precision. Aiming at this problem, we propose a new optimized RBF neural network algorithm based on genetic algorithm (GA-RBF algorithm), which uses genetic algorithm to optimize the weights and structure of RBF neural network; it chooses new ways of hybrid encoding and optimizing simultaneously. Using the binary encoding encodes the number of the hidden layer's neurons and using real encoding encodes the connection weights. Hidden layer neurons number and connection weights are optimized simultaneously in the new algorithm. However, the connection weights optimization is not complete; we need to use least mean square (LMS) algorithm for further leaning, and finally get a new algorithm model. Using two UCI standard data sets to test the new algorithm, the results show that the new algorithm improves the operating efficiency in dealing with complex problems and also improves the recognition precision, which proves that the new algorithm is valid.