Intelligent Data Analysis: An International Journal

Based on the shape characteristics of the sample distribution in the clustering problem, this paper proposes an extended clustering algorithm based on cluster shape boundary (ECBSB). The algorithm automatically determines the number of clusters and classification discrimination boundaries by finding the boundary closures of the clusters from a global perspective of the sample distribution. Since ECBSB is insensitive to local features of the sample distribution, it can accurately identify clusters on complex shape and uneven density distribution. ECBSB first detects the shape boundary points of the cluster in the sample set with edge noise points eliminated, and then generates boundary closures around the cluster based on the boundary points. Finally, the cluster labels of the boundary are propagated to the entire sample set by a nearest neighbor search. The proposed method is evaluated on multiple benchmark datasets. Exhaustive experimental results show that the proposed method achieves highly accurate and robust clustering results, and is superior to the classical clustering baselines on most of the test data.

Classification of imbalanced data is an important challenge in current research. Sampling is an important way to solve the problem of imbalanced data classification, but some traditional sampling algorithms are susceptible to outliers. Therefore, an iF-ADASYN sampling algorithm is proposed in this paper. First, based on the ADASYN algorithm, we introduce the isolation Forest algorithm to overcome its vulnerability to outliers. Then, a calculation method of anomaly index which can delete outliers accurately of minority data is presented. The experimental results of four UCI public imbalanced datasets show that the algorithm can effectively improve the accuracy of the minority class, and increase the stability. In the real thrombus dataset, the AUC value of the iF-ADASYN algorithm is more significant than that of SMOTE and ADASYN algorithms, and the recognition rate of patients with thrombosis increased by 20%. The iF-ADASYN algorithm obtains better resistance to outliers than the original ADASYN algorithm. Meanwhile, it improves the accuracy of minority class decision boundary region division.

The imbalanced data problem is widespread in the real world. In the process of training machine learning models, ignoring imbalanced data problems will cause the performance of the model to deteriorate. At present, researchers have proposed many methods to deal with the imbalanced data problems, but these methods mainly focus on the imbalanced data problems in two-class classification tasks. Learning from multi-class imbalanced data sets is still an open problem. In this paper, an ensemble method for classifying multi-class imbalanced data sets is put forward, called multi-class WHMBoost. It is an extension of WHMBoost that we proposed earlier. We do not use the algorithm used in WHMBoost to process the data, but use random balance based on average size so as to balance the data distribution. The weak classifiers we use in the boosting algorithm are support vector machine and decision tree classifier. In the process of training the model, they participate in training with given weights in order to complement each other’s advantages. On 18 multi-class imbalanced data sets, we compared the performance of multi-class WHMBoost with state of the art ensemble algorithms using MAUC, MG-mean and MMCC as evaluation criteria. The results demonstrate that it has obvious advantages compared with state of the art ensemble algorithms and can effectively deal with multi-class imbalanced data sets.

We consider the problem of clustering a set of uncertain data, where each data consists of a point-set indicating its possible locations. The objective is to identify the representative for each uncertain data and group them into k clusters so as to minimize the total clustering cost. Different from other models, our model does not assume that there is a probability distribution for each uncertain data. Thus, all possible locations need to be considered to determine the representative. Existing methods for this problem are either impractical or have difficulty to handle large-scale datasets due to their pairwise-distance based global search strategy and expensive optimization computation. In this paper, we propose a novel sparse Non-negative Matrix Factorization (NMF) method which measures the similarity of uncertain data by their most commonly shared features. A divide-and-conquer approach is adopted to remarkably improve the efficiency. A novel diagonal l 0 -constraint and its l 1 relaxation are proposed to overcome the challenge of determining the representatives. We give a detailed analysis to show the correctness of our method, and provide an effective initialization and peeling strategy to enhance the ability of processing large-scale datasets. Experimental results on some benchmark datasets confirm the effectiveness of our method.

Named Entity Recognition (NER) is a challenging learning task of identifying and classifying entity mentions in texts into predefined categories. In recent years, deep learning (DL) methods empowered by distributed representations, such as word- and character-level embeddings, have been employed in NER systems. However, for information extraction in Police narrative reports, the performance of a DL-based NER approach is limited due to the presence of fine-grained ambiguous entities. For example, given the narrative report “Anna stole Ada’s car”, imagine that we intend to identify the VICTIM and the ROBBER, two sub-labels of PERSON. Traditional NER systems have limited performance in categorizing entity labels arranged in a hierarchical structure. Furthermore, it is unfeasible to obtain information from knowledge bases to give a disambiguated meaning between the entity mentions and the actual labels. This information must be extracted directly from the context dependencies. In this paper, we deal with the Hierarchical Entity-Label Disambiguation problem in Police reports without the use of knowledge bases. To tackle such a problem, we present HELD, an ensemble model that combines two components for NER: a BLSTM-CRF architecture and a NER tool. Experiments conducted on a real Police reports dataset show that HELD significantly outperforms baseline approaches.

Multiple Instance Learning (MIL) is a widely studied learning paradigm which arises from real applications. Existing MIL methods have achieved prominent performances under the premise of plenty annotation data. Nevertheless, sufficient labeled data is often unattainable due to the high labeling cost. For example, the task in web image identification is to find similar samples among a large size of unlabeled dataset through a small number of provided target pictures. This leads to a particular scenario of Multiple Instance Learning with insufficient Positive and superabundant Unlabeled data (PU-MIL), which is a hot research topic in MIL recently. In this paper, we propose a novel method called Multiple Instance Learning with Bi-level Embedding (MILBLE) to tackle PU-MIL problem. Unlike other PU-MIL method using only simple single-level mapping, the bi-level embedding strategy are designed to customize specific mapping for positive and unlabeled data. It ensures the characteristics of key instance are not erased. Moreover, the weighting measure adopted in positive data can extracts the uncontaminated information of true positive instances without interference from negative ones. Finally, we minimize the classification error loss of mapped examples based on class-prior probability to train the optimal classifier. Experimental results show that our method has better performance than other state-of-the-art methods.

The Internet is a vital part of today’s ecosystem. The speedy evolution of the Internet has brought up practical issues such as the problem of information retrieval. Several methods have been proposed to solve this issue. Such approaches retrieve the information by using SPARQL queries over the Resource Description Framework (RDF) content which requires a precise match concerning the query structure and the RDF content. In this work, we propose a transfer learning-based neural learning method that helps to search RDF graphs to provide probabilistic reasoning between the queries and their results. The problem is formulated as a classification task where RDF graphs are preprocessed to abstract the N-Triples, then encode the abstracted N-triples into a transitional state that is suitable for neural transfer learning. Next, we fine-tune the neural learner to learn the semantic relationships between the N-triples. To validate the proposed approach, we employ ten-fold cross-validation. The results have shown that the anticipated approach is accurate by acquiring the average accuracy, recall, precision, and f-measure. The achieved scores are 97.52%, 96.31%, 98.45%, and 97.37%, respectively, and outperforms the baseline approaches.

Access to one of the richest data sources in the world, the web, is not possible without cost. Often, this cost is not taken into account in data acquisition processes. In this paper, we introduce the Learning Agents (LA) method for automatic topical data acquisition from the web with minimum bandwidth usage and the lowest cost. The proposed LA method uses online learning topical crawlers. The online learning capability makes the LA able to dynamically adapt to the properties of web pages during the crawling process of the target topic, and learn an effective combination of a set of link scoring criteria for that topic. That way, the LA resolves the challenge in the mechanism of combining the outputs of different criteria for computing the value of following a link, in the formerly approaches, and increases the efficiency of the crawlers. A version of the LA method is implemented that uses a collection of topical content analyzers for scoring the links. The learning ability in the implemented LA resolves the challenge of the unclear appropriate size of link contexts for pages of different topics. Using standard metrics in empirical evaluation indicates that when non-learning methods show inefficiency, the learning capability of LA significantly increases the efficiency of topical crawling, and achieves the state of the art results.

In this study, we propose a deep learning related framework to analyze S&P500 stocks using bi-dimensional histogram and autoencoder. The bi-dimensional histogram consisting of daily returns of stock price and stock trading volume is plotted for each stock. Autoencoder is applied to the bi-dimensional histogram to reduce data dimension and extract meaningful features of a stock. The histogram distance matrix for stocks are made of the extracted features of stocks, and stock market network is built by applying Planar Maximally Filtered Graph(PMFG) algorithm to the histogram distance matrix. The constructed stock market network represents the latent space of bi-dimensional histogram, and network analysis is performed to investigate the structural properties of the stock market. we discover that the structural properties of stock market network are related to the dispersion of bi-dimensional histogram. Also, we confirm that the autoencoder is effective in extracting the latent feature of the bi-dimensional histogram. Portfolios using the features of bi-dimensional histogram network are constructed and their investment performance is evaluated in comparison with other benchmark portfolios. We observe that the portfolio consisting of stocks corresponding to the peripheral nodes of bi-dimensional histogram network shows better investment performance than other benchmark stock portfolios.

Existing correlation processing strategies make up for the defect that most evaluation algorithms do not consider the independence between indicators. However, these solutions may change the indicator system’s internal connection, affecting the final evaluation result’s interpretability and accuracy. Besides, traditional independent analysis methods cannot accurately describe the complex multivariate correlation based on the linear relationship. Aimed at these problems, we propose an indicators correlation elimination algorithm based on the feedforward neural network and Taylor expansion (NNTE). Firstly, we propose a generalized n-power correlation and a feedforward neural network to express the relationship between indicators quantitatively. Secondly, the low-order Taylor expression expanded at every sample is pointed to eliminate nonlinear relationships. Finally, to control the expansions’ accuracy, the layer-by-layer stripping method is presented to reduce the dimensionality of the correlations among multiple indicators gradually. This procedure continues to iterate until there are all simple two-dimensional correlations, eliminating multiple variables’ correlations. To compare the elimination efficiency, the ranking accuracy is proposed to measure the distance of the resulting sequence to the benchmark sequence. Under Cleveland and KDD99 two datasets, the ranking accuracy of the NNTE method is 71.64% and 96.41%, respectively. Compared with other seven common elimination methods, our proposed method’s average increase is 13.67% and 25.13%, respectively.

Compact data models have become relevant due to the massive, ever-increasing generation of data. We propose Observers-based Data Modeling (ODM), a lightweight algorithm to extract low density data models (aka coresets) that are suitable for both static and stream data analysis. ODM coresets keep data internal structures while alleviating computational costs of machine learning during evaluation phases accounting for a O(n log n) worst-case complexity. We compare ODM with previous proposals in classification, clustering, and outlier detection. Results show the preponderance of ODM for obtaining the best trade-off in accuracy, versatility, and speed.

Ensemble learners and deep neural networks are state-of-the-art schemes for classification applications. However, deep networks suffer from complex structure, need large amount of samples and also require plenty of time to be converged. In contrast, ensemble learners (especially AdaBoost) are fast to be trained, can work with small and large datasets and also benefit strong mathematical background. In this paper, we have developed a new orthogonal version of AdaBoost, termed as ORBoost, in order to desensitize its performance against noisy samples as well as exploiting low number of weak learners. In ORBoost, after reweighting the distribution of each learner, the Gram-Schmidt rule updates those weights to make a new samples’ distribution to be orthogonal to the former distributions. In contrast in AdaBoost, there is no orthogonality constraint even between two successive weak learners while there is a similarity between the distributions of samples in different learners. To assess the performance of ORBoost, 16 UCI-Repository datasets along with six big datasets are deployed. The performance of ORBoost is compared to the standard AdaBoost, LogitBoost and AveBoost-II over the selected datasets. The achieved results support the significant superiority of ORBoost to the counterparts in terms of accuracy, robustness, number of exploited weak learners and generalization on most of the datasets.