Search

The classification of time series is a central problem in a wide range of disciplines. In this field, the state-of-the-art algorithm is COTE (Collective of Transformation-Based Ensembles) which is a combination of classifiers of different domains: time, autocorrelation , power spectrum and shapelets. The weakest point of this approach is its high computational burden which prevents its use in massive data environments. Shapelet Transform is one of the multiple algorithms that compose this ensemble. It has been shown to achieve a good performance over many reference datasets. Nevertheless, its computational complexity is also too high to be used in massive data environments. On the other hand, Big Data has emerged as an approach to manage massive datasets , which also applies to time series. We propose an algorithm for time series classification in a Big Data environment , DFST. It is based on a combination of the FastShapelet and Shapelet Transform ideas and it is the first completely scalable algorithm for time series classification. We have shown that our proposal scales linearly with the number of time series in dataset. In addition, the classification accuracy is equal to or higher than that of comparable sequential algorithms.

This research has been partially funded by the following grants: TIN2016-81113-R from the Spanish Ministry of Economy and Competitiveness, and P12-TIC-2958 from Andalusian Regional Government, Spain. Francisco J. Baldan holds the FPI grant BES-2017-080137 from the Spanish Ministry of Economy and Competitiveness., Multivariate time series classification is a machine learning task with increasing importance due to the proliferation of information sources in different domains (economy, health, energy, crops, etc.). Univariate methods lack the ability to capture the relationships between the different variables that compose a multivariate time series and therefore cannot be directly extrapolated to multivariate environments. Despite the good performance and competitive results of the multivariate proposals published to date, they are hard to interpret due to their high complexity. In this paper, we propose a multivariate time series classification method based on an alternative representation of the time series, composed of a set of 41 descriptive time series features, in order to improve the interpretability of time series and results obtained. Our proposal uses traditional classifiers over the extracted features to look for relationships between the different variables that form a multivariate time series. We have selected four state-of-the-art algorithms as base classifiers to evaluate our method. We have tested our proposal on the complete University of East Anglia repository, obtaining highly interpretable results capable of explaining the relationships between the features that compose the time series and achieving performance results statistically indistinguishable from the best algorithms of the state-of-the-art., Spanish Ministry of Economy and Competitiveness TIN2016-81113-R, Andalusian Regional Government, Spain P12-TIC-2958, FPI from the Spanish Ministry of Economy and Competitiveness BES-2017-080137

Monitoring complex industrial plants is a very important task in order to ensure the management, reliability, safety and maintenance of the desired product quality. Early detection of abnormal events allows actions to prevent more serious consequences, improve the system's performance and reduce manufacturing costs. In this work, a new methodology for fault detection is introduced, based on time series models and statistical process control (MSPC). The proposal explicitly accounts for both dynamic and non-linearity properties of the system. A dynamic feature selection is carried out to interpret the dynamic relations by characterizing the auto- and cross-correlations for every variable. After that, a time-series based model framework is used to obtain and validate the best descriptive model of the plant (either linear o non-linear). Fault detection is based on finding anomalies in the temporal residual signals obtained from the models by univariate and multivariate statistical process control charts. Finally, the performance of the method is validated on two benchmarks, a wastewater treatment plant and the Tennessee Eastman Plant. A comparison with other classical methods clearly demonstrates the over performance and feasibility of the proposed monitoring scheme.

Cloud Computing is an essential paradigm of computing services based on the “elasticity” property, where available resources are adapted efficiently to different workloads over time. In elastic platforms, the forecasting component can be considered by far the most important element and the differentiating factor when comparing such systems, with workload forecasting one of the problems to solve if we want to achieve a truly elastic system. When properly addressed the cloud workload forecasting problem becomes a really interesting case study. As there is no general methodology in the literature that addresses this problem analytically and from a time series forecasting perspective (even less so in the cloud field), we propose a combination of these tools based on a state-of-the-art forecasting methodology which we have enhanced with some elements, such as: a specific cost function, statistical tests, visual analysis, etc. The insights obtained from this analysis are used to detect the asymmetrical nature of the forecasting problem and to find the best forecasting model from the viewpoint of the current state of the art in time series forecasting. From an operational point of view the most interesting forecast is a short-time horizon, so we focus on this. To show the feasibility of this methodology, we apply it to several realistic workload datasets from different datacenters. The results indicate that the analyzed series are non-linear in nature and that no seasonal patterns can be found. Moreover, on the analyzed datasets, the penalty cost as usually included in the SLA can be reduced to a 30 percent on average.

With the advent of extremely high dimensional datasets, dimensionality reduction techniques are becoming mandatory. Of the many techniques available, feature selection (FS) is of growing interest for its ability to identify both relevant features and frequently repeated instances in huge datasets. We aim to demonstrate that standard FS methods can be parallelized in big data platforms like Apache Spark so as to boost both performance and accuracy. We propose a distributed implementation of a generic FS framework that includes a broad group of well-known information theory-based methods. Experimental results for a broad set of real-world datasets show that our distributed framework is capable of rapidly dealing with ultrahigh-dimensional datasets as well as those with a huge number of samples, outperforming the sequential version in all the cases studied.

Mining massive and high-speed data streams among the main contemporary challenges in machine learning. This calls for methods displaying a high computational efficacy, with ability to continuously update their structure and handle ever-arriving big number of instances. In this paper, we present a new incremental and distributed classifier based on the popular nearest neighbor algorithm, adapted to such a demanding scenario. This method, implemented in Apache Spark, includes a distributed metric-space ordering to perform faster searches. Additionally, we propose an efficient incremental instance selection method for massive data streams that continuously update and remove outdated examples from the case-base. This alleviates the high computational requirements of the original classifier, thus making it suitable for the considered problem. Experimental study conducted on a set of real-life massive data streams proves the usefulness of the proposed solution and shows that we are able to provide the first efficient nearest neighbor solution for high-speed big and streaming data.

In this paper we propose a design for the kNN rule classifier based on GPU devices.This design overcomes the dependency between dataset size and GPU memory required.An efficient CPU-GPU communication scheme is designed.The carefully designed kernels for each step of the computation improve the run-time performance.An efficient selection algorithm replaces the sorting algorithm traditionally used to find the neighborhood. The k nearest neighbor (kNN) rule is one of the most used techniques in data mining and pattern recognition due to its simplicity and low identification error. However, the computational effort it requires is directly related to the dataset sizes, hence delivering a poor performance on large datasets. The use of graphics processing units (GPU) has improved the run-time performance of the kNN rule but the computational requirements of current approaches limit this performance as the dataset size increases.In this paper, we propose a new scalable and memory efficient design for a GPU-based kNN rule, called GPU-SME-kNN, that breaks the dependency between dataset size and memory footprint while delivering high performance. An experimental study of GPU-SME-kNN is presented showing a high performance, even in cases that other methods cannot address, while the computational requirements are suitable for most commercial GPU devices. Our design has also been applied to kNN-based lazy learning algorithms reducing run-times in a significant way.

With the advent of large-scale problems, feature selection has become a fundamental preprocessing step to reduce input dimensionality. The minimum-redundancy-maximum-relevance (mRMR) selector is considered one of the most relevant methods for dimensionality reduction due to its high accuracy. However, it is a computationally expensive technique, sharply affected by the number of features. This paper presents fast-mRMR, an extension of mRMR, which tries to overcome this computational burden. Associated with fast-mRMR, we include a package with three implementations of this algorithm in several platforms, namely, CPU for sequential execution, GPU (graphics processing units) for parallel computing, and Apache Spark for distributed computing using big data technologies.

Discretization is one of the most relevant techniques for data preprocessing. The main goal of discretization is to transform numerical attributes into discrete ones to help the experts to understand the data more easily, and it also provides the possibility to use some learning algorithms which require discrete data as input, such as Bayesian or rule learning. We focus our attention on handling multivariate classification problems, where high interactions among multiple attributes exist. In this paper, we propose the use of evolutionary algorithms to select a subset of cut points that defines the best possible discretization scheme of a data set using a wrapper fitness function. We also incorporate a reduction mechanism to successfully manage the multivariate approach on large data sets. Our method has been compared with the best state-of-the-art discretizers on 45 real datasets. The experiments show that our proposed algorithm overcomes the rest of the methods producing competitive discretization schemes in terms of accuracy, for C4.5, Naive Bayes, PART, and PrUning and BuiLding Integrated in Classification classifiers; and obtained far simpler solutions.

Positive unlabeled time series classification has become an important area during the last decade, as often vast amounts of unlabeled time series data are available but obtaining the corresponding labels is difficult. In this situation, positive unlabeled learning is a suitable option to mitigate the lack of labeled examples. In particular, self-training is a widely used technique due to its simplicity and adaptability. Within this technique, the stopping criterion, i.e., the decision of when to stop labeling, is a critical part, especially in the positive unlabeled context. We propose a self-training method that follows the positive unlabeled approach for time series classification and a family of parameter-free stopping criteria for this method. Our proposal uses a graphical analysis, applied to the minimum distances obtained by the k-Nearest Neighbor as the base learner, to estimate the class boundary. The proposed method is evaluated in an experimental study involving various time series classification datasets. The results show that our method outperforms the transductive results obtained by previous models.

Fuzzy rule-based systems (FRBSs) are a well-known method family within soft computing. They are based on fuzzy concepts to address complex real-world problems. We present the R package frbs which implements the most widely used FRBS models, namely, Mamdani and Takagi Sugeno Kang (TSK) ones, as well as some common variants. In addition a host of learning methods for FRBSs, where the models are constructed from data, are implemented. In this way, accurate and interpretable systems can be built for data analysis and modeling tasks. In this paper, we also provide some examples on the usage of the package and a comparison with other common classification and regression methods available in R., This work was supported in part by the Spanish Ministry of Science and Innovation (MICINN) under Projects TIN2009-14575, TIN2011-28488, TIN2013-47210-P, and P10-TIC-06858. Bergmeir held a scholarship from the Spanish Ministry of Education (MEC) of the \Programa de Formación del Profesorado Universitario (FPU)".

Fingerprints are widely used in a variety of biometric identification systems. The fingerprint matching process is a processing step whose computational requirements limit the size of the fingerprint database that can be dealt with.Fingerprint matching algorithms based on minutiae are one of the most relevant families of biometric identification techniques. The scalability of these models is determined not only by the number of fingerprints but also the number of minutiae per fingerprint. Therefore, processing millions of fingerprints per second requires being able to process hundreds of millions of minutiae per second.In this paper we present a new design of the minutiae based fingerprint matching algorithm presented by Jiang et al. specifically created for GPU based massively parallel architectures. The parallel design allows speed-up ratios of up to 15 with one GPU compared to multi-threaded CPU implementations, and up to 54 using several GPUs in parallel and fingerprint processing rates of between 300,000 and 1,500,000 fingerprints per second.

Throughout the last years, optimization problems on a large number of variables, sometimes over 1000, are becoming common. Thus, algorithms that can tackle them effectively, both in result quality and run time, are necessary. Among these specific algorithms for high-dimensional problems, memetic algorithms, which are the result of the hybridization of an evolutionary algorithm and a local improvement technique, have arisen as very powerful optimization systems for this type of problems. A very effective algorithm of this kind is the MA-SW-Chains algorithm. On the other hand, general purpose computing using Graphics Processing Units (GPUs) has become a very active field because of the high speed-up ratios that can be obtained when applied to problems that exhibit a high degree of data parallelism. In this work we present a new design of MA-SW-Chains to adapt it to the GPU-based massively parallel architecture. The experiments with the new GPU memetic technique, compared to the original sequential version, prove that the results are obtained with the same quality but with a reduction of the run time of two orders of magnitude. This great improvement in computing time makes our proposal suitable for future optimization problems with a dimensionality several orders of magnitude greater than current high-dimensionality standards (i.e. problems with millions of variables). The remarkable run time reduction comes at the cost of a higher design and implementation overhead compared to the CPU-based single-threaded counterpart.

Fingerprint classification is one of the most common approaches to accelerate the identification in large databases of fingerprints. Fingerprints are grouped into disjoint classes, so that an input fingerprint is compared only with those belonging to the predicted class, reducing the penetration rate of the search. The classification procedure usually starts by the extraction of features from the fingerprint image, frequently based on visual characteristics. In this work, we propose an approach to fingerprint classification using convolutional neural networks, which avoid the necessity of an explicit feature extraction process by incorporating the image processing within the training of the classifier. Furthermore, such an approach is able to predict a class even for low-quality fingerprints that are rejected by commonly used algorithms, such as FingerCode. The study gives special importance to the robustness of the classification for different impressions of the same fingerprint, aiming to minimize the penetration in the database. In our experiments, convolutional neural networks yielded better accuracy and penetration rate than state-of-the-art classifiers based on explicit feature extraction. The tested networks also improved on the runtime, as a result of the joint optimization of both feature extraction and classification., This work was partially supported by the Spanish Ministry of Science and Technology under the project TIN2014-57251-P and the Foundation BBVA project 75/2016 BigDaPTOOLS. Y. Saeys is an ISAC Marylou Ingram Scholar.

An increasing amount of unlabeled time series data available render the semi-supervised paradigm a suitable approach to tackle classification problems with a reduced quantity of labeled data. Self-labeled techniques stand out from semi-supervised classification methods due to their simplicity and the lack of strong assumptions about the distribution of the labeled and unlabeled data. This paper addresses the relevance of these techniques in the time series classification context by means of an empirical study that compares successful self-labeled methods in conjunction with various learning schemes and dissimilarity measures. Our experiments involve 35 time series datasets with different ratios of labeled data, aiming to measure the transductive and inductive classification capabilities of the self-labeled methods studied. The results show that the nearest-neighbor rule is a robust choice for the base classifier. In addition, the amending and multi-classifier self-labeled-based approaches reveal a promising attempt to perform semi-supervised classification in the time series context.

The package RoughSets , written mainly in the R language, provides implementations of methods from the rough set theory (RST) and fuzzy rough set theory (FRST) for data modeling and analysis. It considers not only fundamental concepts (e.g., indiscernibility relations, lower/upper approximations, etc.), but also their applications in many tasks: discretization, feature selection, instance selection, rule induction, and nearest neighbor-based classifiers. The package architecture and examples are presented in order to introduce it to researchers and practitioners. Researchers can build new models by defining custom functions as parameters, and practitioners are able to perform analysis and prediction of their data using available algorithms. Additionally, we provide a review and comparison of well-known software packages. Overall, our package should be considered as an alternative software library for analyzing data based on RST and FRST.

In this age, big data applications are increasingly becoming the main focus of attention because of the enormous increment of data generation and storage that has taken place in the last years. This situation becomes a challenge when huge amounts of data are processed to extract knowledge because the data mining techniques are not adapted to the new space and time requirements. Furthermore, real-world data applications usually present a class distribution where the samples that belong to one class, which is precisely the main interest, are hugely outnumbered by the samples of the other classes. This circumstance, known as the class imbalance problem, complicates the learning process as the standard learning techniques do not correctly address this situation.In this work, we analyse the performance of several techniques used to deal with imbalanced datasets in the big data scenario using the Random Forest classifier. Specifically, oversampling, undersampling and cost-sensitive learning have been adapted to big data using MapReduce so that these techniques are able to manage datasets as large as needed providing the necessary support to correctly identify the underrepresented class. The Random Forest classifier provides a solid basis for the comparison because of its performance, robustness and versatility.An experimental study is carried out to evaluate the performance of the diverse algorithms considered. The results obtained show that there is not an approach to imbalanced big data classification that outperforms the others for all the data considered when using Random Forest. Moreover, even for the same type of problem, the best performing method is dependent on the number of mappers selected to run the experiments. In most of the cases, when the number of splits is increased, an improvement in the running times can be observed, however, this progress in times is obtained at the expense of a slight drop in the accuracy performance obtained. This decrement in the performance is related to the lack of density problem, which is evaluated in this work from the imbalanced data point of view, as this issue degrades the performance of classifiers in the imbalanced scenario more severely than in standard learning.

Microarray data classification is a difficult challenge for machine learning researchers due to its high number of features and the small sample sizes. Feature selection has been soon considered a de facto standard in this field since its introduction, and a huge number of feature selection methods were utilized trying to reduce the input dimensionality while improving the classification performance. This paper is devoted to reviewing the most up-to-date feature selection methods developed in this field and the microarray databases most frequently used in the literature. We also make the interested reader aware of the problematic of data characteristics in this domain, such as the imbalance of the data, their complexity, or the so-called dataset shift. Finally, an experimental evaluation on the most representative datasets using well-known feature selection methods is presented, bearing in mind that the aim is not to provide the best feature selection method, but to facilitate their comparative study by the research community.

Fingerprint matching has emerged as an effective tool for human recognition due to the uniqueness, universality and invariability of fingerprints. Many different approaches have been proposed in the literature to determine faithfully if two fingerprint images belong to the same person. Among them, minutiae-based matchers highlight as the most relevant techniques because of their discriminative capabilities, providing precise results. However, performing a fingerprint identification over a large database can be an inefficient task due to the lack of scalability and high computing times of fingerprint matching algorithms.In this paper, we propose a distributed framework for fingerprint matching to tackle large databases in a reasonable time. It provides a general scheme for any kind of matcher, so that its precision is preserved and its time of response can be reduced.To test the proposed system, we conduct an extensive study that involves both synthetic and captured fingerprint databases, which have different characteristics, analyzing the performance of three well-known minutiae-based matchers within the designed framework. With the available hardware resources, our distributed model is able to address up to 400000 fingerprints in approximately half a second. Additional details are provided at http://sci2s.ugr.es/ParallelMatching. HighlightsA two-level parallel AFIS is proposed to deal with large databases.It makes possible to perform pattern identifications in arbitrarily large databases.The framework is flexible for any kind of databases, and any matching algorithm.The achieved speedup is nearly linear.The framework performs 400000 matchings in 0.5s, with no precision loss.

Supported by the Research Projects TIN2014-57251-P, P10-TIC-6858, P12-TIC-2958, TIN2013-47210-P and P11-TIC-7765. I. Triguero holds a BOF postdoctoral fellowship from the Ghent University., The application of data mining and machine learning techniques to biological and biomedicine data continues to be an ubiquitous research theme in current bioinformatics. The rapid advances in biotechnology are allowing us to obtain and store large quantities of data about cells, proteins, genes, etc., that should be processed. Moreover, in many of these problems such as contact map prediction, the problem tackled in this paper, it is difficult to collect representative positive examples. Learning under these circumstances, known as imbalanced big data classification, may not be straightforward for most of the standard machine learning methods. In this work we describe the methodology that won the ECBDL’14 big data challenge for a bioinformatics big data problem. This algorithm, named as ROSEFW-RF, is based on several MapReduce approaches to (1) balance the classes distribution through random oversampling, (2) detect the most relevant features via an evolutionary feature weighting process and a threshold to choose them, (3) build an appropriate Random Forest model from the pre-processed data and finally (4) classify the test data. Across the paper, we detail and analyze the decisions made during the competition showing an extensive experimental study that characterize the way of working of our methodology. From this analysis we can conclude that this approach is very suitable to tackle large-scale bioinformatics classifications problems., Ghent University, TIN2014-57251-P, P10-TIC-6858, P12-TIC-2958, TIN2013-47210-P, P11-TIC-7765

Fingerprint recognition, and in particular minutiae-based matching methods, are ever more deeply implanted into many companies and institutions. As the size of their identification databases grows, there is a need of flexible, reliable structures for fingerprint recognition systems. In this paper, we propose a generic decomposition methodology for minutiae-based matching algorithms that splits the calculation of the matching scores into lower level steps that can be carried out in parallel in a flexible manner. The decomposition allows to adapt any minutiae-based algorithm to frameworks such as MapReduce or Apache Spark. General and specific guidelines to enhance the performance of the adapted matching algorithms are also described. The proposal is evaluated over two matching algorithms, two Big Data frameworks (Apache Hadoop and Apache Spark) and two large-scale fingerprint databases, with promising results concerning the identification time, in addition to the reliability, scalability, distribution and availability capabilities that are provided by such underlying frameworks., TIN2014-57251-P, TIN2013-47210-P, P12-TIC-2958

Fingerprint recognition has been a hot research topic along the last few decades, with many applications and ever growing populations to identify. The need of flexible, fast identification systems is therefore patent in such situations. In this context, fingerprint classification is commonly used to improve the speed of the identification. This paper proposes a complete identification system with a hierarchical classification framework that fuses the information of multiple feature extractors. A feature selection is applied to improve the classification accuracy. Finally, the distributed identification is carried out with an incremental search, exploring the classes according to the probability order given by the classifier. A single parameter tunes the trade-off between identification time and accuracy. The proposal is evaluated over two NIST databases and a large synthetic database, yielding penetration rates close to the optimal values that can be reached with classification, leading to low identification times with small or no accuracy loss., TIN2014-57251-P, TIN2013-47210-P, P12-TIC-2958

Global optimization is an important field of research both in mathematics and computer sciences. It has applications in nearly all fields of modern science and engineering. Memetic algorithms are powerful problem solvers in the domain of continuous optimization, as they offer a trade-off between exploration of the search space using an evolutionary algorithm scheme, and focused exploitation of promising regions with a local search algorithm. In particular, we describe the memetic algorithms with local search chains (MA-LS-Chains) paradigm, and the R package Rmalschains, which implements them. MA-LS-Chains has proven to be effective compared to other algorithms, especially in high-dimensional problem solving. In an experimental study, we demonstrate the advantages of using Rmalschains for high-dimension optimization problems in comparison to other optimization methods already available in R., This work was supported in part by the Spanish Ministry of Science and Innovation (MICINN) under Project TIN-2009-14575. The work was performed while C. Bergmeir held a scholarship from the Spanish Ministry of Education (MEC) of the “Programa de Formación del Profesorado Universitario (FPU)”.

This work was supported by the research projects TIN2014-57251-P, TIN2013-47210-P and P12-TIC-2958. D. Peralta holds an FPU scholarship from the Spanish Ministry of Education and Science (FPU12/04902). I. Triguero holds a BOF postdoctoral fellowship from the Ghent University., Nowadays, many companies and institutions need fast and reliable identification systems that are able to deal with very large databases. Fingerprints are among the most used biometric traits for identification. In the current literature there are fingerprint matching algorithms that are focused on efficiency, whilst others are based on accuracy. In this paper we propose a flexible dual phase identification method, called DPD-DFF, that combines two fingers and two matchers within a hybrid fusion scheme to obtain both fast and accurate results. Different alternatives are designed to find a trade-off between runtime and accuracy that can be further tuned with a single parameter. The experiments show that DPD-DFF obtains very competitive results in comparison with the state-of-the-art score fusion techniques, especially when dealing with large databases or impostor fingerprints., Spanish Government FPU12/04902, Ghent University, TIN2014-57251-P, TIN2013-47210-P, P12-TIC-2958

The main objective of "time series analysis" is to discover the underlying structure of the time series, and thus, become able to forecast its "future values". This process makes it possible to predict, control or simulate variables. Most of the time series modelling procedures try to forecast future values from lagged ones. Thus, the selection of the relevant lagged values to be used is a key step. In this paper, a new consensus method for the selection of relevant lagged values of a time series is introduced: feature ranking aggregated selection (FRASel). The main contribution of this feature selection method is the definition of a consensus decision making mechanism based on aggregation and expressed as a simple rule. In FRASel, the selected subset of lagged values is decided by the application of an aggregation criterion to the results of different flavours of feature ranking methods, applied from different approaches. A thorough empirical analysis is carried out to assess the performance of FRASel. The statistical significance of the experimental results is also analysed through the application of non-parametric statistical tests.

In this brief, we present a novel model fitting procedure for the neuro-coefficient smooth transition autoregressive model (NCSTAR), as presented by Medeiros and Veiga. The model is endowed with a statistically founded iterative building procedure and can be interpreted in terms of fuzzy rule-based systems. The interpretability of the generated models and a mathematically sound building procedure are two very important properties of forecasting models. The model fitting procedure employed by the original NCSTAR is a combination of initial parameter estimation by a grid search procedure with a traditional local search algorithm. We propose a different fitting procedure, using a memetic algorithm, in order to obtain more accurate models. An empirical evaluation of the method is performed, applying it to various real-world time series originating from three forecasting competitions. The results indicate that we can significantly enhance the accuracy of the models, making them competitive to models commonly used in the field.

In time series predictor evaluation, we observe that with respect to the model selection procedure there is a gap between evaluation of traditional forecasting procedures, on the one hand, and evaluation of machine learning techniques on the other hand. In traditional forecasting, it is common practice to reserve a part from the end of each time series for testing, and to use the rest of the series for training. Thus it is not made full use of the data, but theoretical problems with respect to temporal evolutionary effects and dependencies within the data as well as practical problems regarding missing values are eliminated. On the other hand, when evaluating machine learning and other regression methods used for time series forecasting, often cross-validation is used for evaluation, paying little attention to the fact that those theoretical problems invalidate the fundamental assumptions of cross-validation. To close this gap and examine the consequences of different model selection procedures in practice, we have developed a rigorous and extensive empirical study. Six different model selection procedures, based on (i) cross-validation and (ii) evaluation using the series' last part, are used to assess the performance of four machine learning and other regression techniques on synthetic and real-world time series. No practical consequences of the theoretical flaws were found during our study, but the use of cross-validation techniques led to a more robust model selection. To make use of the ''best of both worlds'', we suggest that the use of a blocked form of cross-validation for time series evaluation became the standard procedure, thus using all available information and circumventing the theoretical problems.

Soft computing (SC) emerged as an integrating framework for a number of techniques that could complement one another quite well (artificial neural networks, fuzzy systems, evolutionary algorithms, probabilistic reasoning). Since its inception, a distinctive goal has been to dig out the deep relationships among their components. This paper considers two wide families of SC models. On the one hand, the regime-switching autoregressive paradigm is a recent development in statistical time series modeling, and it includes a set of models closely related to artificial neural networks. On the other hand, we consider fuzzy rule-based systems in the framework of time series analysis. This paper discloses original results establishing functional equivalences between models of these two classes, and hence opens the door to a productive line of research where results and techniques from one area can be applied in the other. As a consequence of the equivalences presented in this paper, we prove the asymptotic stationarity of a class of fuzzy rule-based systems. Simulations based on information criteria show the importance of the selection of the proper membership function.

The use of feature selection can improve accuracy, efficiency, applicability and understandability of a learning process. For this reason, many methods of automatic feature selection have been developed. Some of these methods are based on the search of the features that allows the data set to be considered consistent. In a search problem we usually evaluate the search states, in the case of feature selection we measure the possible feature sets. This paper reviews the state of the art of consistency based feature selection methods, identifying the measures used for feature sets. An in-deep study of these measures is conducted, including the definition of a new measure necessary for completeness. After that, we perform an empirical evaluation of the measures comparing them with the highly reputed wrapper approach. Consistency measures achieve similar results to those of the wrapper approach with much better efficiency.

This work was supported by the Research Projects CAB (CDTI), TIN2011-28488, and TIN2013-40765-P. D. Peralta holds an FPU scholarship from the Spanish Ministry of Education and Science (FPU12/04902)., Fingerprint recognition has found a reliable application for verification or identification of people in biometrics. Globally, fingerprints can be viewed as valuable traits due to several perceptions observed by the experts; such as the distinctiveness and the permanence on humans and the performance in real applications. Among the main stages of fingerprint recognition, the automated matching phase has received much attention from the early years up to nowadays. This paper is devoted to review and categorize the vast number of fingerprint matching methods proposed in the specialized literature. In particular, we focus on local minutiae-based matching algorithms, which provide good performance with an excellent trade-off between efficacy and efficiency. We identify the main properties and differences of existing methods. Then, we include an experimental evaluation involving the most representative local minutiae-based matching models in both verification and evaluation tasks. The results obtained will be discussed in detail, supporting the description of future directions., Research Projects CAB (CDTI) TIN2011-28488 TIN2013-40765-P, Spanish Government FPU12/04902

This paper reviews the fingerprint classification literature looking at the problem from a double perspective. We first deal with feature extraction methods, including the different models considered for singular point detection and for orientation map extraction. Then, we focus on the different learning models considered to build the classifiers used to label new fingerprints. Taxonomies and classifications for the feature extraction, singular point detection, orientation extraction and learning methods are presented. A critical view of the existing literature have led us to present a discussion on the existing methods and their drawbacks such as difficulty in their reimplementation, lack of details or major differences in their evaluations procedures. On this account, an experimental analysis of the most relevant methods is carried out in the second part of this paper, and a new method based on their combination is presented., Research Projects CAB(CDTI) TIN2011-28488 TIN2013-40765, Spanish Government FPU12/04902

In the first part of this paper we reviewed the fingerprint classification literature from two different perspectives: the feature extraction and the classifier learning. Aiming at answering the question of which among the reviewed methods would perform better in a real implementation we end up in a discussion which showed the difficulty in answering this question. No previous comparison exists in the literature and comparisons among papers are done with different experimental frameworks. Moreover, the difficulty in implementing published methods was stated due to the lack of details in their description, parameters and the fact that no source code is shared. For this reason, in this paper we will go through a deep experimental study following the proposed double perspective. In order to do so, we have carefully implemented some of the most relevant feature extraction methods according to the explanations found in the corresponding papers and we have tested their performance with different classifiers, including those specific proposals made by the authors. Our aim is to develop an objective experimental study in a common framework, which has not been done before and which can serve as a baseline for future works on the topic. This way, we will not only test their quality, but their reusability by other researchers and will be able to indicate which proposals could be considered for future developments. Furthermore, we will show that combining different feature extraction models in an ensemble can lead to a superior performance, significantly increasing the results obtained by individual models., Research Projects CAB(CDTI) TIN2011-28488 TIN2013-40765, Spanish Government FPU12/04902

Learning from data is a process to construct a model according to available training data so that it can be used to make predictions for new data. Nowadays, several software libraries are available to carry out this task, frbs is an R package which is aimed to construct models from data based on fuzzy rule based systems (FRBSs) by employing learning procedures from Computational Intelligence (e.g., neural networks and genetic algorithms) to tackle classification and regression problems. For the learning process, frbs considers well-known methods, such as Wang and Mendel's technique, ANFIS, Hy-FIS, DENFIS, subtractive clustering, SLAVE, and several others. Many options are available to perform conjunction, disjunction, and implication operators, defuzzification methods, and membership functions (e.g., triangle, trapezoid, Gaussian, etc). It has been developed in the R language which is an open-source analysis environment for scientific computing. In this paper, we also provide some examples on the usage of the package and a comparison with other software libraries implementing FRBSs. We conclude that frbs should be considered as an alternative software library for learning from data.

Neural networks are important standard machine learning procedures for classification and regression. We describe the R package RSNNS that provides a convenient interface to the popular Stuttgart Neural Network Simulator SNNS. The main features are (a) encapsulation of the relevant SNNS parts in a C++ class, for sequential and parallel usage of different networks, (b) accessibility of all of the SNNS algorithmic functionality from R using a low-level interface, and (c) a high-level interface for convenient, R-style usage of many standard neural network procedures. The package also includes functions for visualization and analysis of the models and the training procedures, as well as functions for data input/output from/to the original SNNS file formats., This work was supported in part by the Spanish Ministry of Science and Innovation (MICINN) under Project TIN-2009-14575. C. Bergmeir holds a scholarship from the Spanish Ministry of Education (MEC) of the \Programa de Formación del Profesorado Universitario (FPU)".

Highlights? A system for semi-automatic processing of full-resolution whole-slide scans.? A new algorithm for segmenting the nuclei under control of the expert user.? Data storage and interaction of technical and medical experts is facilitated.? Open source software with web-based architecture.. In order to automate cervical cancer screening tests, one of the most important and longstanding challenges is the segmentation of cell nuclei in the stained specimens. Though nuclei of isolated cells in high-quality acquisitions often are easy to segment, the problem lies in the segmentation of large numbers of nuclei with various characteristics under differing acquisition conditions in high-resolution scans of the complete microscope slides. We implemented a system that enables processing of full resolution images, and proposes a new algorithm for segmenting the nuclei under adequate control of the expert user. The system can work automatically or interactively guided, to allow for segmentation within the whole range of slide and image characteristics. It facilitates data storage and interaction of technical and medical experts, especially with its web-based architecture. The proposed algorithm localizes cell nuclei using a voting scheme and prior knowledge, before it determines the exact shape of the nuclei by means of an elastic segmentation algorithm. After noise removal with a mean-shift and a median filtering takes place, edges are extracted with a Canny edge detection algorithm. Motivated by the observation that cell nuclei are surrounded by cytoplasm and their shape is roughly elliptical, edges adjacent to the background are removed. A randomized Hough transform for ellipses finds candidate nuclei, which are then processed by a level set algorithm. The algorithm is tested and compared to other algorithms on a database containing 207 images acquired from two different microscope slides, with promising results.

In time series analysis remaining autocorrelation in the errors of a model implies that it is failing to properly capture the structure of time-dependence of the series under study. This can be used as a diagnostic checking tool and as an indicator of the adequacy of the model. Through the study of the errors of the model in the Lagrange Multiplier testing framework, in this paper we derive (and validate using simulated and real world examples) a hypothesis test which allows us to determine if there is some left autocorrelation in the error series. This represents a new diagnostic checking tool for fuzzy rule-based modelling of time series and is an important step towards statistically sound modelling strategy for fuzzy rule-based models.

This paper presents an extension of the method presented by Benitez et al (1997) for extracting fuzzy rules from an artificial neural network (ANN) that express exactly its behavior. The extraction process provides an interpretation of the ANN in terms of fuzzy rules. The fuzzy rules presented are in accordance with the domain of the input variables. These rules use a new operator in the antecedent. The properties and intuitive meaning of this operator are studied. Next, the role of the biases in the fuzzy rule-based systems is analyzed. Several examples are presented to comment on the obtained fuzzy rule-based systems. Finally, the interpretation of ANNs with two or more hidden layers is also studied.