Search

We submitted this paper to the special issue, "Four Decades of Rough Set Theory: Achievements and Future." We have researched a theory and an execution tool for rule generation from a Deterministic Information System (DIS) and a Non-deterministic Information System (NIS). We developed the NIS-Apriori algorithm, which combines the rough sets-based concept and the Apriori algorithm, for rule generation from NIS. We term this research series as a Rough set Non-deterministic Information Analysis (RNIA). In the first half of this paper, we describe the framework of RNIA and its execution environment, as well as the results we achieved. Then, later in this paper, we enumerate various applications of RNIA, such as detection of data dependencies, decision support, estimation and completion of missing values, the problem of learning DIS from NIS, and generation of rules from non-tabular and multiple heterogeneous data sets. They are our current and prospective subjects. RNIA's capabilities can lead to several developments. • Our past research on RNIA is reviewed with its execution videos. • Attribute dependency (or feature selection) analysis using the obtained rules is studied. • Solutions to Missing value imputation and Machine Learning by Rule Generation (MLRG) are presented with their execution videos. • A Descriptor-based Information System (DbIS) for rule generation from non-tabular data sets is studied. [ABSTRACT FROM AUTHOR]

[Display omitted] • New technique on recurrent neural network models to predict the conditions in hybrid-digital model of cardiovascular system. • Novel approach to combination of data from a laboratory model and neural network models to correlate the parameters of the cardiovascular system. • Presentation of possible application of the unique hybrid-digital model of cardiovascular system. End-systolic elastance of the left ventricle along with the waveforms of pressure and volumetric blood flow in particular sectors of the circulatory system are of importance in diagnosing various problems like dilated cardiomyopathy, left-ventricular hypertrophy, pulmonary hypertension, or ischemic heart disease. The objective of the paper is to broaden the spectrum of available methods to estimate those parameters since currently accessible techniques are often costly or troublesome. Six models have been developed − three of them estimate end-systolic elastance, two perform regression of volumetric blood flow, and one predicts blood pressure. Training datasets have been collected applying the unique hybrid-digital model. The input of the designed models consists of two or three different waveforms representing pressure and volumetric blood flow in particular areas, including heart ventricles, atria, and pulmonary vessels, in addition to the heart rate value. The basis of each model comprises bidirectional Long Short-Term Memory layers along with the dropout and feed-forward layers. Models that estimate end-systolic elastance achieved various accuracy. One of them performed exceptionally well since the absolute error did not exceed 0.169 m m H g c m 3 which is a negligibly small value. The root-mean-square error (RMSE) of the model predicting pressure waveform reached 0.165 m m H g in the worst case. Regression of the volumetric blood flow resulted in 6.062 c m 3 s worst-case RMSE for the model focusing on the pulmonary valve and 15.979 c m 3 s for pulmonary veins model. Computed results, especially those of the models estimating end-systolic elastance, indicate that it is possible to utilize neural networks to estimate those parameters with sufficient accuracy. [ABSTRACT FROM AUTHOR]

Zircon U-Pb SIMS dating combined with in-context (in thin section) monazite and xenotime U + Th-total Pb dating was used to clarify the Palaeozoic evolution of the 'cold' Chopok granite (Nízke Tatry Mountains, Slovakia). Four distinct zircon, monazite and xenotime age domains testify to a prolonged evolution from igneous formation to multi-stage metasomatism and hydrothermal overprinting. The geological interpretation of age patterns from 'cold' granites, expected to have low zircon saturation temperatures (<800 °C) and relatively high amounts of zircon inheritance, requires special care, especially for what concerns proper attribution of zircon inheritance and igneous growth ages. These issues can be resolved using zircon saturation temperatures (T Zrn) as proxy for the amount of zircon inheritance in combination with the temperature differences between T Zrn and the granite solidus. In this respect, the Chopok granite is an atypical 'cold' granite. Due to T Zrn being substantially lower (ca. 80 °C ± 50 °C) than the granite solidus temperature, practically no zircon inheritance was found. The zircon age data indicates that the Chopok granite is a product of an Early Ordovician (475.8 ± 3.3 Ma) magmatic event, corresponding with the widespread Early Palaeozoic magmatism recorded throughout the European Variscan belt. This is further corroborated by phosphate mineral ages. The post-magmatic activity recorded in the U-Pb systematics of zircon and phosphates overgrowths can be related to the different phases of the evolution of the Variscan orogen: Early Carboniferous (ca. 352 Ma) metasomatism documents the main Variscan orogenic event, whereas the Permo-Triassic age (ca. 255 Ma) reflects thermo-tectonic activity associated with large-scale crustal extension, contemporaneous with the initial continental leading to the break-up of Pangea. Unlabelled Image • 'Cold' - 'hot' granite classification is not ubiquitously valid. • The Chopok granite is an atypical 'cold' granite in showing little inheritance. • The Chopok granite formation is Early Ordovician not Carboniferous. [ABSTRACT FROM AUTHOR]

Machine learning methods for the identification of child sexual abuse materials (CSAM) have been previously studied, however, they have serious limitations. Firstly, the training sets used to train the appropriate machine learning algorithms were not previously annotated by a forensic expert in anthropology. Secondly, previously presented solutions have rarely used models trained using real pornographic content involving children. Thirdly, previous studies have not presented a detailed justification for the classification decisions made, which is important due to the recent guidelines of the European Commission (Artificial Intelligence Act). The aim of the study was to train convolution neural networks (CNNs) using expert-labelled CSAM images and thereby identify the elements of the body and/or the environment that are critical for classifications by the neural network. To train and evaluate machine learning models, we used 60,000 images equally divided into four classes (CSAM images, images displaying sexual activity involving adults, images of people without sexual activity, and images not containing people). We used four neural network architectures: MobileNet, ResNet152, xResNet152 and its modification ResNet-s, designed for the purpose of research. The trained models provided high accuracy of classifying CSAM images: xResNet152 (F1 = 0.93, 92,8%), xResNet-s (F1 = 0.93, 93,1%), ResNet152 (F1 = 0.90, 91,39%), MobileNet (F1 ranged from 0.85 to 0.87, accuracy ranged from 86% to 87%). The results of the conducted research suggest that using expert knowledge (in sexology and anthropology) significantly improved the accuracy of the models. In regard to further anthropological analysis, the results indicate that the breasts, face and torso are crucial areas for the classification of pornographic content with children's participation. Results suggests that the ResNet-s neural network may be a reliable tool for clinical work and to support the work of experts witnesses in the field of anthropology. The study design received a positive opinion of the Ethics Committee of the Faculty of Mathematics, Informatics and Mechanics of the University of Warsaw. The clinical material was used for research purposes with the consent of the relevant prosecutor's offices. Authors provided free version of Windows application to classify CSAM for forensic experts, policemen and prosecutors at the OSF repository (DOI: 10.17605/OSF.IO/RU7JX). • Machine learning is believed to be the best solution for CSAM identification. • The machine learning models cannot replace human, but it can assist their work. • Neural networks are not fully explainable. • Using expert knowledge improves the explainability of the machine learning models. [ABSTRACT FROM AUTHOR]

The paper presents a novel approach to investigating mistakes in machine learning model operations. The considered approach is the basis for BrightBox – a diagnostic technology that can be used for analyzing prediction models and identifying model- and data-related issues. The idea is to generate surrogate rough set-based models from data that approximate decisions made by monitored black-box models. Such approximators are used to compute neighborhoods of instances that undergo the diagnostic process — the neighborhoods consist of historical instances that were processed in a similar way by rough set-based models. The diagnostic process is then based on the analysis of mistakes registered in such neighborhoods. The experiments performed on real-world data sets confirm that such analysis can provide us with efficient and valid insights about the reasons for the poor performance of machine learning models. • BrightBox uses rough set-based approximators to construct surrogates of complex black-box models. • We defined a neighborhood of an instance based on a similarity in the decision-making process. • BrightBox computes diagnostic attributes to characterize operations of diagnosed prediction models. • Aggregations of diagnostic attributes allow the assessment of the model's generalization capabilities. [ABSTRACT FROM AUTHOR]

Nowadays high strength steels are gained increasing popularity for use in steel constructions and machinery components. Their fatigue properties, however, are not yet well characterized, especially in comparison with widespread conventional structural steel S355J2. In this study were conducted the assessment of use HSS steel in extremely loaded components with existing plastic strains. In order to obtain the necessary characteristics and fatigue properties, materials were tested under LCF strain-based mode with strain ratio R ɛ = 0,1 and strain rate of 10 -2 1/s. Because of its higher strength, S960QL steel exhibits a slightly higher fatigue resistant under the total strain amplitude lower than 0,5% compared to mild steel and similar under higher strains. Obtained results show that investigated HSS steel can successfully supersede presently used conventional mild steel in heavily loaded components simultaneously decreasing dead weight of the structure. [ABSTRACT FROM AUTHOR]

Due to the constant increase of energy demand, new ecological sources of energy are being sought. One of the promising directions is the use of algae biomass from special cultivation or biomass floating on the sea. In order to determine the potential of algae for thermal decomposition, the pyrolysis study in thermobalance coupled with mass spectrometer was carried out at heating rates of 5, 10, 20 °C/min, in the temperature range of 30–900 °C Three types of algae: macroalgae Cladophora glomerata and Fucus vesiculosus , and microalgae Chlorella pyrenoidesa were used in this research. During the deconvolution of the derivative of the thermogravimetric curves, three peaks were assigned to the decomposition of carbohydrates, proteins and lipids. The isoconversional kinetic model used indicated that the pyrolysis process occurred in several reactions. The proposed kinetic model consists of three nth order independent parallel reactions of pseudo-components. The calculated activation energy ranged from 39 to 266 kJ/mol for carbohydrates, 177–181 kJ/mol for proteins, and 110–229 kJ/mol for lipids. The obtained results could be used in the design of 3rd generation biorefineries. • Chlorella pyrenoidesa , Cladophora glomerata , Fucus vesiculosus were pyrolysed. • The greatest weight loss of algae in pyrolysis was in range temperature 200–600 °C. • Evolved gasses (H 2 , CH 4 , H 2 O, CO, CO 2) from the pyrolysis process were identified. • Deconvolution of DTG curves determines algae composition. • The proposed kinetic model consists of three parallel reactions. [ABSTRACT FROM AUTHOR]

Background The purpose of this study was to evaluate the efficacy of arthroscopic elbow release for both traumatic and degenerative contractures from intraoperative recording through the recovery time until final follow-up. Methods The study is based on 54 consecutive patients with extrinsic elbow contracture (traumatic in 31 and degenerative in 23) treated with arthroscopic arthrolysis by a single surgeon in 2011-2015. Range of motion (ROM) and the Mayo Elbow Performance Score (MEPS) were recorded preoperatively; intraoperatively; following release; and in the 1st, 3rd, 8th, 12th, and 26th weeks and at 2 years postoperatively. Results Significant improvements were noted in extension, flexion, and range of motion, measured both intraoperatively and at all follow-up visits. The greatest improvement in the range of motion was achieved at the time of surgery (from 89° ± 28° to 131° ± 14°, P  < .001); it then decreased at 1 week to 103° ± 22° ( P  < .001) and slowly recovered to reach 124° ± 22° after 2 years. This was better than the preoperative value ( P  < .001) but worse than the intraoperative value ( P  = .002). A similar pattern was observed in both traumatic and degenerative contractures. The MEPS improved from 73 ± 12 preoperatively to 93 ± 14 at the final evaluation ( P  < .001). The ROM and MEPS results at every follow-up were comparable for both traumatic and degenerative contractures. ROM improved regardless of the severity of contracture. Conclusions Arthroscopic elbow arthrolysis was similarly efficient in ROM restoration in both traumatic and degenerative contractures and regardless of the severity of contracture. After early deterioration, the achieved gain slowly recovers over a period of 6 months but may not recover to the ranges achieved during arthroscopy. [ABSTRACT FROM AUTHOR]