Search

Abstract The study explore machine learning (ML) techniques to predict temperature-dependent photoluminescence (PL) spectra in colloidal CdSe nanoplatelets (NPLs), leveraging polynomial regression models trained on experimental data from 85 to 270 K spanning temperatures to forecast PL spectra backward to 0 K and forward to 300 K. 6th-degree polynomial models with Tweedie regression were optimal for band energy ( $$B_1$$ ) predictions up to 300 K, while 9th-degree models with LassoLars and Linear Regression regressors were suitable for backward predictions to 0 K. For exciton energy ( $$B_2$$ ), the Lasso model of degree 5 and the Ridge model of degree 4 performed well up to 300 K, while the Tweedie model of degree 2 and Theil-Sen model of degree 2 showed promise for predictions to 0 K. Furthermore, a GA-based approach was utilized to fit experimental data to theoretical model of Fan and Varshni equations, facilitating a comparative analysis with the ML-predicted curves.

This study presents an integrated approach for automatically extracting and structuring information from medical reports, captured as scanned documents or photographs, through a combination of image recognition and natural language processing (NLP) techniques like named entity recognition (NER). The primary aim was to develop an adaptive model for efficient text extraction from medical report images. This involved utilizing a genetic algorithm (GA) to fine-tune optical character recognition (OCR) hyperparameters, ensuring maximal text extraction length, followed by NER processing to categorize the extracted information into required entities, adjusting parameters if entities were not correctly extracted based on manual annotations. Despite the diverse formats of medical report images in the dataset, all in Russian, this serves as a conceptual example of information extraction (IE) that can be easily extended to other languages.

In recent years, cybersecurity management has increasingly required advanced methodologies capable of handling complex, evolving threat landscapes. Scenario network-based approaches have emerged as effective strategies for managing uncertainty and adaptability in cybersecurity projects. This article introduces a scenario network-based approach for managing cybersecurity projects, utilizing fuzzy linguistic models and a Takagi–Sugeno–Kanga fuzzy neural network. Drawing upon L. Zadeh’s theory of linguistic variables, the methodology integrates expert analysis, linguistic variables, and a continuous genetic algorithm to predict membership function parameters. Fuzzy production rules are employed for decision-making, while the Mamdani fuzzy inference algorithm enhances interpretability. This approach enables multi-scenario planning and adaptability across multi-stage cybersecurity projects. Preliminary results from a research prototype of an intelligent expert system—designed to analyze project stages and adaptively construct project trajectories—suggest the proposed approach is effective. In computational experiments, the use of fuzzy procedures resulted in an over 25% reduction in errors compared to traditional methods, particularly in adjusting project scenarios from pessimistic to baseline projections. While promising, this approach requires further testing across diverse cybersecurity contexts. Future studies will aim to refine scenario adaptation and optimize system response in high-risk project environments.

This study explores the potential application of the mechanical activation (MA) of nickel powder for incorporation into the composition of powder wire blends for the deposition of wear-resistant coatings. Nickel powder of PNE-1 grade was processed in a vibrational mill for various durations (4 to 16 min) with different combinations of grinding media. The influence of MA parameters on the bulk density and apparent particle size of nickel powder was investigated. The greatest effect was observed at the maximum processing time of 16 min, where electron microscopy revealed significant deformation and an increase in discoid particles, leading to enhanced energy accumulation. Nickel powder processed with a combination of 6 balls that are 20 mm in diameter and 8 balls that are 10 mm in diameter showed significant changes, though no major alteration in chemical composition was noted. XRMA indicated that the powder’s surface was partially covered with oxides, with a composition of 96.8–98.4% Ni and 0.8–1.7% O2. Additionally, the effect of nickel powders after the treatment on the structure of deposited metal was determined, demonstrating alterations in the morphology and a slight increase in hardness. Furthermore, a convolutional neural network (CNN)-based approach was proposed to discern fragments within images depicting surface microstructures, both with and without MA.

The influence of acoustic anisotropy on ultrasonic testing reliability poses a challenge in evaluating products from additive technologies (AT). This study investigates how elasticity constants of anisotropic materials affect defect signal amplitudes in AT products. Experimental measurements on AT samples were conducted to determine elasticity constants. Using Computational Modeling and Simulation Software (CIVA), simulations explored echo signal changes across ultrasound propagation directions. The parameters A13 (the ratio between the velocities of ultrasonic transverse waves with vertical and horizontal polarizations at a 45-degree angle to the growth direction), A3 (the ratio for waves at a 90-degree angle), and Ag (the modulus of the difference between A13 and A3) were derived from wave velocity relationships and used to characterize acoustic anisotropy. Comparative analysis revealed a strong correlation (0.97) between the proposed anisotropy coefficient Ag and the amplitude changes. Threshold values of Ag were introduced to classify anisotropic materials based on observed amplitude changes in defect echo signals. In addition, a method leveraging deep learning to predict Ag based on data from other anisotropy constants through genetic algorithm (GA)-optimized neural network (NN) architectures is proposed, offering an approach that can reduce the computational costs associated with calculating such constants.

This study investigates the methods for controlling porosity in thermal pipes manufactured using selective laser melting (SLM) technology. Experiments conducted include water permeability tests and surface roughness measurements, which are complemented by SEM image ML-based analysis for pore recognition. The results elucidate the impact of SLM printing parameters on water permeability. Specifically, an increase in hatch and point distances leads to a linear rise in permeability, while higher laser power diminishes permeability. Using machine learning (ML) techniques, precise pore identification on SEM images depicting surface microstructures of the samples is achieved. The average percentage of the surface area containing detected pores for microstructure samples printed with laser parameters (laser power (W) _ hatch distance (µm) _ point distance (µm)) 175_ 80_80 was found to be 5.2%, while for 225_120_120, it was 4.2%, and for 275_160_160, it was 3.8%. Pore recognition was conducted using the Haar feature-based method, and the optimal patch size was determined to be 36 pixels on monochrome images of microstructures with a magnification of 33×, which were acquired using a Leica S9 D microscope.

This study proposes a method for classifying economic activity descriptors to match Nomenclature of Economic Activities (NACE) codes, employing a blend of machine learning techniques and expert evaluation. By leveraging natural language processing (NLP) methods to vectorize activity descriptors and utilizing genetic algorithm (GA) optimization to fine-tune hyperparameters in multi-class classifiers like Naive Bayes, Decision Trees, Random Forests, and Multilayer Perceptrons, our aim is to boost the accuracy and reliability of an economic classification system. This system faces challenges due to the absence of precise target labels in the dataset. Hence, it is essential to initially check the accuracy of utilized methods based on expert evaluations using a small dataset before generalizing to a larger one.

The paper introduces a computer vision methodology for detecting pitting corrosion in gas pipelines. To achieve this, a dataset comprising 576,000 images of pipelines with and without pitting corrosion was curated. A custom-designed and optimized convolutional neural network (CNN) was employed for binary classification, distinguishing between corroded and non-corroded images. This CNN architecture, despite having relatively few parameters compared to existing CNN classifiers, achieved a notably high classification accuracy of 98.44%. The proposed CNN outperformed many contemporary classifiers in its efficacy. By leveraging deep learning, this approach effectively eliminates the need for manual inspection of pipelines for pitting corrosion, thus streamlining what was previously a time-consuming and cost-ineffective process.