Search

Nowadays, both predictive and descriptive modelling play a key role in decision-making processes in almost every branch of activity. In this article we are introducing $IntelliDaM$ , a generic machine learning-based framework useful for improving the performance of data mining tasks and subsequently enhancing decision-making processes. Through its components designed for feature analysis, unsupervised and supervised learning-based data mining, $IntelliDaM$ facilitates hidden knowledge discovery from data. Intensive research has been conducted in the field of educational data mining, as education institutions are interested in constantly adapting their educational programs to the needs of society by improving the quality of managerial decisions, course instructors’ decision-making, or information gathering for course design. The present work conducts a longitudinal educational data mining study by applying $IntelliDaM$ to real data collected at Babeş-Bolyai University, Romania, for a Computer Science course. The problem of mining educational data has been thoroughly examined using the proposed framework, with the goal of analysing students’ performance. A very good performance has been achieved for the classification task (an $F1$ score of around 92%), and the results also highlighted a statistically significant performance improvement by using a technique for selecting discriminative data features. The performed study confirmed that $IntelliDaM$ could be a useful instrument in educational environments, particularly for improving decision-making processes, like designing courses, the setup of efficient examinations, avoiding plagiarism, or offering support regarding stress management.

Software Defects Prediction represents an essential activity during software development that contributes to continuously improving software quality and software maintenance and evolution by detecting defect-prone modules in new versions of a software system. In this paper, we are conducting an in-depth analysis on the software features’ impact on the performance of deep learning-based software defect predictors. We further extend a large-scale feature set proposed in the literature for detecting defect-proneness, by adding conceptual software features that capture the semantics of the source code, including comments. The conceptual features are automatically engineered using Doc2Vec, an artificial neural network based prediction model. A broad evaluation performed on the Calcite software system highlights a statistically significant improvement obtained by applying deep learning-based classifiers for detecting software defects when using conceptual features extracted from the source code for characterizing the software entities.

Recently proposed improvements in the field of Computer Vision refer to enhancing the feature processing capabilities of Single-Task Convolutional Neural Networks. A typical Single-Task network consists of a backbone and a head, where the feature extractor is usually optimised using the gradient provided by the head. Inevitably, the backbone specialises for the given task. This sort of approach does not scale well for learning multiple tasks at once while having the same input. As a response, there is an increasing interest in Multi-Task formulations. Since most Multi-Task architectures employ a single shared backbone, when gradients from different tasks are propagated back to it, it can result in its oversaturation. Thus, this problem may be solved using Multi-Backbone feature extractors. Hence, as a strategy proposed to compensate for these shortcomings, we introduce MBMT-Net, a Multi-Backbone-Multi-Task-Network architecture based on a development strategy that infuses backbones with more diverse and specialised processing capabilities. MBMT-Net consists of parallel pre-trained backbones whose outputs are concatenated and offered to the Multi-Task heads that shall benefit from richer and more diverse features with decreased number of network parameters when compared to traditional Multi-Task architectures. Our strategy is architecture independent, and it can be applied to different types of backbones and parsing heads, which greatly extends the domain of configurable features, finally enhancing existing Single- and Multi-Task model building strategies and outperforming them when using the Multi-Backbone design. As a result, while having a deficit of 12.16M parameters, MBMT-Net reaches state-of-the-art performances, and surpasses the previously best semantic segmentation Multi-Task model in terms of Mean Intersection over Union when evaluated on NYUv2 data set.

Flash floods are a major weather-related risk, as they cause more than 5000 fatalities annually, according to the World Meteorological Organization. Quantitative Precipitation Estimation is a method used to approximate the rainfall over locations where direct field observations are not available. It represents one of the most valuable information employed by meteorologists and hydrologists for issuing early warnings concerning flash floods. The current study is in line with the efforts to improve radar-based rainfall estimates through the use of machine learning techniques applied on radar data. With this aim, as a proof of concept, six machine learning models are evaluated to make estimations of the radar-based hourly accumulated rainfall using reflectivity data collected on the lowest radar elevation angles, and we employ a new data model for representing these radar data. The data were collected by a WSR-98D weather radar of the Romanian Meteorological Administration, located in the central region of Romania, during 30 non-consecutive days of the convective seasons, between 2016 and 2021. We obtained encouraging results using a stacked machine learning model. In terms of the Root Mean Squared Error evaluation metric, the results of the proposed stacked regressor are better than the radar estimated accumulated rainfall by about 33% and also outperform the baseline computed using the Z-R relationship by about 13%.

With the recent increase in the occurrence of severe weather phenomena, the development of accurate weather nowcasting is of paramount importance. Among the computational methods that are used to predict the evolution of weather, deep learning techniques offer a particularly appealing solution due to their capability for learning patterns from large amounts of data and their fast inference times. In this paper, we propose a convolutional network for weather forecasting that is based on radar product prediction. Our model (NeXtNow) adapts the ResNeXt architecture that has been proposed in the computer vision literature to solve the spatiotemporal prediction problem. NeXtNow consists of an encoder–decoder convolutional architecture, which maps radar measurements from the past onto radar measurements that are recorded in the future. The ResNeXt architecture was chosen as the basis for our network due to its flexibility, which allows for the design of models that can be customized for specific tasks by stacking multiple blocks of the same type. We validated our approach using radar data that were collected from the Romanian National Meteorological Administration (NMA) and the Norwegian Meteorological Institute (MET) and we empirically showed that the inclusion of multiple past radar measurements led to more accurate predictions further in the future. We also showed that NeXtNow could outperform XNow, which is a convolutional architecture that has previously been proposed for short-term radar data prediction and has a performance that is comparable to those of other similar approaches in the nowcasting literature. Compared to XNow, NeXtNow provided improvements to the critical success index that ranged from 1% to 17% and improvements to the root mean square error that ranged from 5% to 6%.

Software authorship attribution, defined as the problem of software authentication and resolution of source code ownership, is of major relevance in the software engineering field. Authorship analysis of source code is more difficult than the classic task on literature, but it would be of great use in various software development activities such as software maintenance, software quality analysis or project management. This paper addresses the problem of code authorship attribution and introduces, as a proof of concept, a new supervised classification model AutoSoft for identifying the developer of a certain piece of code. The proposed model is composed of an ensemble of autoencoders that are trained to encode and recognize the programming style of software developers. An extension of the AutoSoft classifier, able to recognize an unknown developer (a developer that was not seen during the training), is also discussed and evaluated. Experiments conducted on software programs collected from the Google Code Jam data set highlight the performance of the proposed model in various test settings. A comparison to existing similar solutions for code authorship attribution indicates that AutoSoft outperforms most of them. Moreover, AutoSoft provides the advantage of adaptability, illustrated through a series of extensions such as the definition of class membership probabilities and the re-framing of the AutoSoft system to address one-class classification.

Understanding students’ learning processes and education-related phenomena by extracting knowledge from educational data sets represents a continuous interest in the educational data mining domain. Due to an accelerated expansion of online learning and digitalisation in education, there is a growing interest in understanding the impact of online learning on the academic performance of students. In this study, we comparatively investigate traditional and synchronous online learning methods to assess students’ performance through the use of deep autoencoders. Experiments performed on real data sets collected in both online and traditional learning environments showed that autoencoders are able to detect hidden patterns in academic data sets unsupervised; these patterns are valuable for the prediction of students’ performance. The obtained results emphasized that, for the considered case studies, traditional evaluations are a little more accurate than online evaluations. Still, after applying a one-tailed paired Wilcoxon signed-rank test, no statistically significant difference between the traditional and online evaluations was observed.

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.

Proteins are essential molecules, that must correctly perform their roles for the good health of living organisms. The majority of proteins operate in complexes and the way they interact has pivotal influence on the proper functioning of such organisms. In this study we address the problem of protein–protein interaction and we propose and investigate a method based on the use of an ensemble of autoencoders. Our approach, entitled AutoPPI, adopts a strategy based on two autoencoders, one for each type of interactions (positive and negative) and we advance three types of neural network architectures for the autoencoders. Experiments were performed on several data sets comprising proteins from four different species. The results indicate good performances of our proposed model, with accuracy and AUC values of over 0.97 in all cases. The best performing model relies on a Siamese architecture in both the encoder and the decoder, which advantageously captures common features in protein pairs. Comparisons with other machine learning techniques applied for the same problem prove that AutoPPI outperforms most of its contenders, for the considered data sets.

One of the hottest topics in today’s meteorological research is Weather nowcasting, which is the weather forecast for a short time period such as one to six hours. Radar is an important data source used by operational meteorologists for issuing nowcasting warnings. With the main goal of helping meteorologists in analysing radar data for issuing nowcasting warnings, we propose NowDeepN, a supervised learning based regression model which uses an ensemble of deep artificial neural networks for predicting the values for radar products at a certain time moment. The values predicted by NowDeepN may be used by meteorologists in estimating the future development of potential severe phenomena and would replace the time consuming process of extrapolating the radar echoes. NowDeepN is intended to be a proof of concept for the effectiveness of learning from radar data relevant patterns that would be useful for predicting future values for radar products based on their historical values. For assessing the performance of NowDeepN, a set of experiments on real radar data provided by the Romanian National Meteorological Administration is conducted. The impact of a data cleaning step introduced for correcting the erroneous radar products’ values is investigated both from the computational and meteorological perspectives. The experimental results also indicate the relevance of the features considered in the supervised learning task, highlighting that the radar products’ values at a certain geographical location at a time moment may be predicted from the products’ values from a neighboring area of that location at previous time moments. An overall Normalized Root Mean Squared Error less than 4% was obtained for NowDeepN on the cleaned radar data. Compared to similar related work from the nowcasting literature, NowDeepN outperforms several approaches and this emphasizes the performance of our proposal.

The maintenance and evolution of software systems are highly impacted by activities such as bug fixing, adding new features or functionalities and updating existing ones. Impact analysis contributes to improving the maintenance activities by determining those parts from a software system which can be affected by changes to the system. There exist hidden dependencies in the software projects which cannot be found using common coupling measures and are due to the so called indirect coupling. In this paper we aim to provide a comprehensive review of existing methods for hidden dependencies identification, as well as to highlight the limitations of the existing state-of-the-art approaches. We also propose an unsupervised learning based computational model for the problem of hidden dependencies identification and give some incipient experimental results. The study performed in this paper supports our broader goal of developing machine learning methods for automatically detecting hidden dependencies.

Proteins are large, complex molecules with crucial roles in the functioning of living organisms. Understanding the underlying mechanisms by which proteins achieve their structures and substructures, as well as those involved in the conformational transitions may contribute to a deeper comprehension of the involved biological processes. This paper investigates a new machine learning perspective upon analyzing protein conformational transitions and introduces a new formalization for the problem, with the more general goal of uncovering interesting patterns in protein conformational transitions. This study represents the starting point of a research which is being conducted in order to obtain a better comprehension of proteins’ structures and, implicitly, functions, by investigating computational intelligence methods for analyzing and deducing proteins conformational transitions.

In this paper we aim at investigating and experimentally evaluating the reinforcement learning based model that we have previously introduced to solve the well-known matrix bandwidth minimization problem (MBMP). The MBMP is an NP-complete problem, which is to permute rows and columns of a matrix in order to keep its nonzero elements in a band lying as close as possible to the main diagonal. The MBMP has been found to be relevant to a wide range of applications including circuit design, network survivability, data storage and information retrieval. The potential of the reinforcement learning model proposed for solving the MBMP was confirmed by the computational experiment, which has provided encouraging results.

Temporal modeling and analysis and more specifically, temporal ordering are very important problems within the fields of bioinformatics and computational biology, as the temporal analysis of the events characterizing a certain biological process could provide significant insights into its development and progression. Particularly, in the case of cancer, understanding the dynamics and the evolution of this disease could lead to better methods for prediction and treatment. In this paper we tackle, from a computational perspective, the temporal ordering problem, which refers to constructing a sorted collection of multi-dimensional biological data, collection that reflects an accurate temporal evolution of biological systems. We introduce a novel approach, based on reinforcement learning, more precisely, on Q-learning, for the biological temporal ordering problem. The experimental evaluation is performed using several DNA microarray data sets, two of which contain cancer gene expression data. The obtained solutions are correlated either to the given correct ordering (in the cases where this is provided for validation), or to the overall survival time of the patients (in the case of the cancer data sets), thus confirming a good performance of the proposed model and indicating the potential of our proposal.