The classification of variable stars is essential for understanding stellar evolution and dynamics. With the growing volume of light curve data from extensive surveys, there is a need for automated and accurate classification methods. Traditional methods often rely on manual feature extraction and selection, which can be time-consuming and less adaptable to large datasets. In this work, we present an approach using a convolutional neural network (CNN) to classify variable stars using only raw light curve data and their known periods, without the need for manual feature extraction or hand-selected data preprocessing. Our method utilizes phase-folding to organize the light curves and directly learns the variability patterns crucial for classification. Trained and tested on the Optical Gravitational Lensing Experiment (OGLE) dataset, our model demonstrates an average accuracy of 88% and an F1 score of 0.89 across five well-known classes of variable stars. We also compared our classification model with the Random Forest (RF) classifier and showed that our model gives better results across all of the classification metrics. By leveraging CNN, our approach does not need manual feature extraction and can handle diverse light curve shapes and sampling cadences. This automated, data-driven method offers a powerful tool for classifying variable stars, enabling efficient processing of large datasets from current and future sky surveys.
30 September 2024 in Journal of Astronomy and Space Sciences
This study developed a machine learning-based methodology to classify gravitational wave (GW) signals from black hole-neutron star (BH-NS) mergers by combining convolutional neural network (CNN) with conditional information for feature extraction. The model was trained and validated on a dataset of simulated GW signals injected to Gaussian noise to mimic real world signals. We considered all three types of merger: binary black hole (BBH), binary neutron star (BNS) and neutron star-black hole (NSBH). We achieved up to 96% correct classification of GW signals sources. Incorporating our novel conditional information approach improved classification accuracy by 10% compared to standard time series training. Additionally, to show the effectiveness of our method, we tested the model with real GW data from the Gravitational Wave Transient Catalog (GWTC-3) and successfully classified ~90% of signals. These results are an important step towards low-latency real-time GW detection.
Over the past decade, wireless mesh networks (WMNs) have seen significant advancements due to their simple deployment, cost-effectiveness, ease of implementation, and reliable service coverage. However, despite these advantages, the placement of nodes in WMNs presents a critical challenge that significantly impacts their performance. This issue is recognized as an NP-hard problem, underscoring the necessity of development optimization algorithms, such as heuristic and metaheuristic approaches. This motivated us to develop the Maximum Entropy Genetic Algorithm (MEGA) to address the issue of mesh router node placement in WMNs. To assess the proposed method, we conducted experiments across various scenarios with different settings, focusing on key metrics such as network connectivity and user coverage. The simulation results showed the comparative performance of MEGA in relation to other prominent algorithms, such as the Coyote Optimization Algorithm (COA), Firefly Algorithm (FA), Genetic Algorithm (GA), and Particle Swarm Optimization (PSO), revealing MEGA’s effectiveness and usability in determining optimal locations for mesh routers.
30 September 2024 in Eurasian Physical Technical Journal
Quantitative analysis of the structure of star clusters is crucial for understanding their formation and evolution. In this article, we explore the application of fractal dimension analysis to study the evolution of star clusters, also fractal dimension, a concept from fractal geometry, provides a quantitative measure of the complexity and self-similarity of geometric objects. By considering star clusters as complex networks, we employ the box covering method to calculate their fractal dimension. Our methodology combines the well-established Minimum Spanning Tree (MST) and Box-Covering (BC) methods and using these methods, the fractal structure of the clusters was determined. It was revealed that star clusters disintegrate at a fractal dimension of 1.3 and obey a power law. It should be noted that the obtained result was compared with the results of the McLuster.
29 December 2025 in Eurasian Physical Technical Journal
Gamma-ray bursts represent some of the most energetic and complex phenomena in the universe, characterized by highly variable light curves that often contain observational gaps. Reconstructing these light curves is essential for gaining deeper insight into the physical processes driving such events. This study proposes a machine learning-based framework for the reconstruction of gamma-ray burst light curves, focusing specifically on the plateau phase observed in X-ray data. The analysis compares the performance of three sequential modeling approaches: a bidirectional recurrent neural network, a gated recurrent architecture, and a convolutional model designed for temporal data. The findings of this study indicate that the Bidirectional Gated Recurrent Unit model showed the best predictive accuracy among the evaluated models across all gamma-ray burst types, as measured by Mean Absolute Error, Root Mean Square Error, and Coefficient of Determination. Notably, Bidirectional Gated Recurrent Unit exhibited enhanced capability in modeling both gradual plateau phases and abrupt transient features, including flares and breaks, particularly in complex light-curve scenarios.