Search

The paper approaches the topic of nowcasting, one of the hottest topics in meteorology which deals with the problem of short-term forecasting of severe weather phenomena. Various types of meteorological data, including radar measurements, satellite data and weather stations’ observations are currently used for forecasting severe weather events. Radar data is one of the important sources used by meteorologists for nowcasting and for providing alerts for severe weather events. We are proposing a new one-class classifier, named RadRAR (Radar products’ values prediction using Relational Association Rules) for convective storms nowcasting based on radar data. More specifically, RadRAR is trained on radar data collected from normal weather conditions and learns to predict whether the radar echo values will be higher than 35dBZ, i.e. likely to indicate the occurrence of a storm. RadRAR is intended to be a proof a concept that relational association rule mining applied on radar data is helpful in discriminating between severe and normal weather conditions. Real radar data provided by the Romanian National Meteorological Administration is used for evaluating the effectiveness of RadRAR. A Critical Success Index of 61% was obtained, outperforming similar approaches from the literature and highlighting a good performance of RadRAR.

Identifying defective software components is an essential activity during software development which contributes to continuously improving the software quality. Since relatively numerous defects are due to violated software dependencies, coupling metrics could increase the performance of software defect prediction. Among various measures expressing the coupling between software components, the conceptual coupling metrics capture similarities based on the semantic information contained in the source code. We are introducing a new conceptual coupling based metric suite, named COMET, for software defect prediction. Experiments conducted on publicly available data sets, using both unsupervised and supervised learning models, emphasize that COMET metrics suite is superior to the software metrics widely used in the defect prediction literature.

Data mining techniques are intensively used to uncover relevant patterns in large volumes of complex data which are continuously extended with newly arrived data instances. Relational association rules (RARs), a data analysis and mining concept, have been introduced as an extension of classical association rules (ARs) for capturing various relationships between the attributes characterizing the data. Due to its NP-completeness, the problem of mining all the interesting RARs within a data set is computationally difficult. As the dimensionality of the data set to be mined increases, the classical algorithm Discovery of Relational Association Rules (DRAR) for RARs mining fails in providing the set of rules in reasonable time. This paper introduces a new approach named CRAR (Concurrent Relational Association Rule mining) which uses concurrency for the RARs discovery process and thus significantly reduces the mining time. The effectiveness of CRAR is empirically validated on nine open source data sets. The reduction in mining time when using CRAR against DRAR emphasizes that it can be successfully applied in various practical data mining scenarios.

Coupling is a fundamental property of software systems which is strongly connected with the quality of software design and has high impact on program understanding. The coupling between software components influences software maintenance and evolution as well. In order to ease the maintenance and evolution processes it is essential to estimate the impact of changes made in the software system, coupling indicating such a possible impact. This paper introduces a new aggregated coupling measurement which captures both the structural and the conceptual characteristics of coupling between the software components. The proposed measure combines the textual information contained in the source code with the structural relationships between software components. We conduct several experiments which underline that the proposed aggregated coupling measure reveals new characteristics of coupling and is also effective for change impact analysis.

The main purpose of the Educational Data Mining domain is to provide additional insights into the students’ learning mechanism and thus to offer a better understanding of the educational processes. This paper investigates the usefulness of unsupervised machine learning methods, particularly principal component analysis and relational association rule mining in analysing students’ academic performance data, with the broader goal of developing supervised learning models for students’ performance prediction. Experiments performed on a real academic data set highlight the potential of unsupervised learning models for uncovering meaningful patterns within educational data, patterns which will be relevant for predicting the students’ academic performance.

The ability to understand and model proteins’ dynamics is of great relevance in biology and medicine. A good comprehension of the way the proteins change their structure is important as these transitions give the function of the protein within the organism. In this paper we are introducing an unsupervised learning based approach using variational autoencoders, for uncovering protein motions and conformational transitions. The main goal of the research is to offer an interpretable method for proteins’ trajectories visualisation by learning a low dimensional space that accurately represents the input data, as empirically confirmed through the performed experiments. An additional aim is to comparatively evaluate the impact of two protein representations on the learning process.

Relational Association Rules (RARs) capture generic relations between attributes values in possibly large data sets. Due to their ability to uncover underlying semantically relevant patterns, they are of particular interest in data mining research and applicable in both unsupervised and supervised learning scenarios. With the aim of increasing the stability and expressiveness of the classical, non-gradual RARs, Gradual Relational Association Rules (GRARs) have been introduced. By generalizing the boolean relations to gradual relations, GRARs also capture the degrees to which generic relations are satisfied. In the current paper we introduce a new approach called DynGRAR (Dynamic Gradual Relational Association Rules Miner) for uncovering interesting GRARs in dynamic data sets which are incrementally extended with both new data instances and new data attributes. DynGRAR dynamically adjusts the set of all interesting GRARs. Through multiple experiments performed on publicly available software defect prediction data sets, we have evaluated DynGRAR versus applying the standard GRARs mining algorithm from scratch on the extended data. The results obtained emphasize the superior performance of the dynamic approach we propose.

This paper analyses the problem of predicting students’ academic performance, a subject that is increasingly investigated within the Educational Data Mining literature. For a better understanding of the educational related phenomena, there is a continuous interest in applying supervised and unsupervised learning methods for obtaining additional insights into the students’ learning process. The problem of predicting if a student will pass or fail at a certain academic discipline based on the students’ grades received during the semester is a difficult one, highly dependent on various conditions such as the course, the number of examinations during the semester, the instructors and their exigences. We propose a new classification model, S PRAR (Students Performance prediction using Relational Association Rules) for predicting the final result of a student at a certain academic discipline using relational association rules (RARs). RARs extend the classical association rules for expressing various relationships between data attributes. Experiments are performed on three real academic data sets collected from Babes-Bolyai University from Romania. The performance of the S PRAR classifier on the considered case studies is compared against existing related work, being superior to previously proposed students’ performance predictors.

Predicting weather, and particularly severe weather, is an important challenge both for meteorological and machine learning researchers. The complexity and difficulty of the problem is mainly due to the chaotic character of the atmosphere and the implicit large set of meteorological information (radar, satellite or ground meteorological observations) which have to be analyzed by meteorologists. Thus, understanding the relationships between various meteorological parameters extracted from radar observations may be useful for providing additional comprehension about severe weather development and would help to identify situations when severe weather can occur. Self-organizing maps are being explored as an unsupervised classification model for detecting patterns in radar data which are relevant in predicting short-term weather changes. Experiments are performed on real radar data provided by the Romanian National Meteorological Administration. With the main goal of analyzing how the values for the weather radar products are evolving between consecutive radar scans, we empirically show that in general there is a slow change in the values over time, except for the situations when certain severe phenomena occur. The study conducted in this paper is aimed to provide a better insight regarding how the values of weather radar products are evolving in time both in calm and severe weather conditions, with the broader goal of using these findings for weather nowcasting.

The growing complexity of software projects requires increasing consideration of their analysis and testing. Identifying defective software entities is essential for software quality assurance and it also improves activities related to software testing. In this study, we developed a novel supervised classification method called HyGRAR for software defect prediction. HyGRAR is a non-linear hybrid model that combines gradual relational association rule mining and artificial neural networks to discriminate between defective and non-defective software entities. Experiments performed based on 10 open-source data sets demonstrated the excellent performance of the HYGRAR classifier. HyGRAR performed better than most of the previously proposed approaches for software defect prediction in performance evaluations using the same data sets.

Within the software testing domain, determining the order in which classes are tested in integration testing is an important problem called class integration test order identification (CITO). This problem is useful in integration testing, as it contributes to reducing the time needed for testing a software system and refers to the process of identifying an optimal order in which the application classes should be combined and tested as a group. This paper introduces a novel approach based on reinforcement learning for class integration test order optimization in the context of integration testing. The experimental evaluation is performed on four synthetic examples and on six existing software systems often used in the literature for this problem. The results obtained are analyzed and compared to similar related work from the literature, highlighting the potential of the current proposal. The proposed reinforcement learning based approach outperforms most methods existing in the software engineering literature for optimizing the test order for class-based integration. Moreover, it is general, and can be easily extended for optimizing the order in which software components should be tested during component integration testing of a software system.

Online data mining techniques are used to uncover relevant patterns in complex data which are dynamic by nature and thus continuously extended with real-time arriving data streams. Relational association rules (RARs), a data analysis and mining concept, extend the classical association rules so as to capture different relations between the attributes characterizing the data. This paper introduces a new Incremental Relational Association Rule Mining (IRARM) approach with the aim of progressively adapting the interesting relational association rules identified in a data set, when it is enlarged with new instances. We have experimentally evaluated IRARM on publicly available data sets. The reduction in mining time when using IRARM against mining from scratch emphasizes its efficiency in adapting the rules to real-time data extension.

Weather nowcasting which is the analysis and shortterm weather forecast is a topic of major interest both for the meteorological and machine learning researchers. The problem is a complex one due to the large volume of data (such as radar, satellite or other ground meteorological observations) which has to be analyzed by meteorologists for issuing nowcasting warnings. In addition, climate changes are chaotic and the climate models are complex. The main goal of the paper is to better understand the relationships between the meteorological products extracted from radar observations both in severe and normal weather conditions. Self organizing maps are proposed as unsupervised learning models for detecting hidden patterns in radar data which are well correlated with weather changes. Experiments performed on real radar data provided by the Romanian National Meteorological Administration highlight the potential of unsupervised learning to uncover in radar data hidden rules which are relevant from a meteorological perspective. The results of our study suggest promising results in applying predictive supervised learning models for weather nowcasting based on radar data.

Short-term quantitative precipitation forecast is a challenging topic in meteorology, as the number of severe meteorological phenomena is increasing in most regions of the world. Weather radar data is of utmost importance to meteorologists for issuing short-term weather forecast and warnings of severe weather phenomena. We are proposing AutoNowP, a binary classification model intended for precipitation nowcasting based on weather radar reflectivity prediction. Specifically, AutoNowP uses two convolutional autoencoders, being trained on radar data collected on both stratiform and convective weather conditions for learning to predict whether the radar reflectivity values will be above or below a certain threshold. AutoNowP is intended to be a proof of concept that autoencoders are useful in distinguishing between convective and stratiform precipitation. Real radar data provided by the Romanian National Meteorological Administration and the Norwegian Meteorological Institute is used for evaluating the effectiveness of AutoNowP. Results showed that AutoNowP surpassed other binary classifiers used in the supervised learning literature in terms of probability of detection and negative predictive value, highlighting its predictive performance.

Proteomics is nowadays one of the most important and relevant fields from computational biology, raising a lot of challenging and provocative questions. Gaining an understanding of protein dynamic and function as well as obtaining additional insights into the protein folding process is still of great interest in bioinformatics and medicine. This paper introduces a new approach A n o m a l P for detecting anomalous protein conformational transitions using deep autoencoders for encoding information about the structural similarity between proteins belonging to the same superfamily. Experiments are conducted on real protein data and the obtained results emphasize the potential of autoencoders to learn biological relevant patterns, such as proteins’ structural characteristics and that they are useful for detecting conformations or proteins which are likely to be anomalous with respect to a superfamily. The study performed in this paper is aimed to provide better insights of proteins structural similarity, with the broader goal of learning to predict proteins conformational transitions.

Software defect prediction has extensive applicability thus being a very active research area in Search-Based Software Engineering. A high proportion of the software defects are caused by violated couplings. In this paper, we investigate the relevance of semantic coupling in assessing the software proneness to defects. We propose a hybrid classification model combining Gradual Relational Association Rules with Artificial Neural Networks, which detects the defective software entities based on semantic features automatically learned from the source code. The experiments we have performed led to results that confirm the interplay between conceptual coupling and software defects proneness.

It is not easy finding arguments against the common belief that Proteomics and Genomics are the most challenging and important research fields, posing interesting problems for our current era. Gaining insight into the protein folding process has been the goal of many in the past few decades. Understanding completely how proteins come alive and behave will revolutionize modern medicine. With the main goal of understanding the importance of the protein folding problem and uncovering hidden patterns in protein data, we are analyzing protein conformational transitions with unsupervised learning tools, by applying different types of hard and fuzzy clustering algorithms and comparing the results. As an additional goal, the paper describes a software that can perform on demand analysis on protein data and display the results in a web interface. It is a proof of concept for potential useful features that make software algorithms available for researchers of all domains.

Predicting the students’ academic achievements is of great interest within the Educational Data Mining (EDM) field, having the main goal of extracting patterns relevant in deciding the students’ performance at a certain academic course. This paper analyses a classification model SPRAR (Students Performance prediction using Relational Association Rules) for predicting the academic results of students using relational association rules. Relational association rules represent an extension of classical association rules able to capture binary relationships between the attribute values. Three new classification scores are introduced in this paper and used in the classification stage of SPRAR. Their performance is analyzed using three real academic data sets from Babes-Bolyai University, Romania and compared to similar existing results from the literature.

Cytogenetics is a field of genetics investigating the relationships between the hereditary characteristics, structure and behavior of human chromosomes, as well as the medical and evolutionary repercussions of chromosomal abnormalities. Detecting the human karyotype and chromosomal anomalies could offer relevant information about human genetics and possible genetic disorders. This paper investigates an automatic solution for chromosomes classification and introduces an unsupervised learning approach KarySOM based on self-organizing maps for the problem of automatically human karyotyping, with the more general goal of uncovering chromosomal anomalies. The experimental evaluation of the proposed approach highlights its effectiveness for unsupervised classification of human chromosomes.

Convolutional neural networks are effective supervised learning models which are widely used nowadays in various applications ranging from computer vision tasks such as image detection and classification, image captioning, to video classification. Even if the convolutional models are highly performant, a major drawback is given by their computationally expensiveness from the viewpoint of the required memory, additions and multiplications operations and thus are hardly portable on limited-resource devices. The purpose of this paper is to demonstrate the applicability of convolutional neural networks for low resource devices and to study their performance in real life scenarios. In this respect, with the major goal of preserving the performance, we propose a convolutional neural network model, called SimpLeNet, using distillation for image tagging that can run on low-resource devices such as smartphones, smartwatches, tablets or TVs. Experiments performed on MNIST data set for image classification emphasize the effectiveness of SimpLeNet, both in terms of model’s size reduction, as well as in terms of classification accuracy

The study of protein dynamics through analysis of conformational transitions represents a significant stage in understanding protein function. Using molecular simulations, large samples of protein transitions can be recorded. However, extracting functional motions from these samples is still not automated and extremely time-consuming. In this paper we investigate the usefulness of unsupervised machine learning methods for uncovering relevant information about protein functional dynamics. Autoencoders are being explored in order to highlight their ability to learn relevant biological patterns, such as structural characteristics. This study is aimed to provide a better comprehension of how protein conformational transitions are evolving in time, within the larger framework of automatically detecting functional motions.

The task of estimating the body mass from human skeletal remains based on bone measurements is an important one in bioarchaeology and forensic anthropology. Most of the current literature deals with this problem through mathematical linear regression formulas applied to various bones. In order to improve the existing results, two supervised learning-based regression models are proposed, using artificial neural networks and support vector machines, which are useful for expressing good (usually nonlinear) mappings between skeletal measurements and body mass. Several experiments performed on an open source data set show that the proposed applications of machine learning-based algorithms lead to better results than the current state of the art. Thus, the proposed methods are useful for producing good body mass estimations from skeletal measurements.

Multiple sequence alignment plays an important role in comparative genomic sequence analysis, being one of the most challenging problems in bioinformatics. This problem refers to the process of arranging the primary sequences of DNA, RNA or protein to identify regions of similarity that may be a consequence of functional, structural or evolutionary relationships between the sequences. In this paper we tackle multiple sequence alignment from a computational perspective and we introduce a novel approach, based on reinforcement learning, for addressing it. The experimental evaluation is performed on several DNA data sets, two of which contain human DNA sequences. The efficiency of our algorithm is shown by the obtained results, which prove that our technique outperforms other methods existing in the literature and which also indicate the potential of our proposal.

More and more sophisticated malware attacks are developed nowadays and new variants of existing malicious software are released daily. Malware clustering is often applied to identify patterns of malicious software, with similar samples being grouped together and considered variants of the same malware family. In this paper we propose an automated technique based on agglomerative hierarchical clustering combined with a supervised learning method for parameters optimization which helps determining samples that exhibit the same behavior, allowing malware analysts to uncover new and interesting threats. The proposed method relies on behavioral and attack pattern analysis. Despite the complexity of nowadays malicious software, the attacks of same malware families are very similar in terms of actions performed on the infected system. The experimental evaluation is carried out on a real case study and the results are analyzed, interpreted and compared to the ones of similar existing approaches. Our experiments demonstrate the capability of the proposed clustering method to accurately identify groups of similar malware samples, which are very likely to represent malware families.

An essential characteristic of a software system, which is strongly related to its quality and has a major impact on program comprehension, is coupling. The coupling between software components also significantly influences the maintenance and evolution. We propose in this paper a coupling measure for Object-Oriented (OO) software systems which quantifies conceptual coupling of application classes. Conceptual coupling measures how the sources of different software entities (components, classes, modules, etc.) relate to each other, considering the textual information (identifiers, comments, etc.) contained in the code. We express the conceptual relationship between two software entities using an unsupervisedly learned high-dimensional representation of their text descriptions. Several experiments are conducted to emphasize that the proposed conceptual coupling measure expresses other aspects of coupling than the structural coupling measures and to also test its effectiveness for software restructuring at package level.

Software refactoring at the package level has a major importance in the field of software architecture, since refactoring increases the internal software quality and is beneficial during software maintenance and evolution. As the requirements for grouping application classes into software packages are hard to identify, clustering is useful, since it is able to uncover hidden patterns in data. In this paper we are investigating software refactoring at the package level by using hierarchical clustering. A hierarchical clustering approach is proposed in order to help software developers in designing well-structured software packages. The approach takes an existing software system and remodularizes it at the package level using hierarchical clustering, in order to obtain better-structured packages. The experimental evaluation of the approach is performed on three open source frameworks and the algorithm has proven to perform well.

Detecting defective entities from existing software systems is a problem of great importance for increasing both the software quality and the efficiency of software testing related activities. We introduce in this paper a novel approach for predicting software defects using fuzzy decision trees. Through the fuzzy approach we aim to better cope with noise and imprecise information. A fuzzy decision tree will be trained to identify if a software module is or not a defective one. Two open source software systems are used for experimentally evaluating our approach. The obtained results highlight that the fuzzy decision tree approach outperforms the non-fuzzy one on almost all case studies used for evaluation. Compared to the approaches used in the literature, the fuzzy decision tree classifier is shown to be more efficient than most of the other machine learning-based classifiers.