Journal of Problems in Computer Science and Information Technologies

AIR QUALITY PREDICTION BASED ON THE LSTM WITH ATTENTION USING METEOROLOGICAL DATA IN URBAN AREA IN KAZAKHSTAN

2025-02-25T03:48:07+00:00

his study investigates air pollution prediction in urban Kazakhstan, specifically focusing on Almaty, utilizing machine learning models, LightGBM, and Long Short-Term Memory (LSTM) with an attention mechanism. The research addresses the limitations of current air quality monitoring systems and aims to improve the accuracy of predicting PM2.5 and PM10 concentrations using meteorological data. Results demonstrated that while LightGBM efficiently handled tabular data, LSTM with attention exhibited predictive accuracy by capturing temporal dependencies and handling data variability more effectively. LSTM with attention achieved RMSE values of 5.54 and 5.69 for PM2.5 and PM10, respectively, compared to LightGBM's 4.75 and 5.76. The findings also highlight correlations between pollution levels and environmental conditions such as time of day, wind direction, and temperature. We conclude that LSTM with attention is better suited for air quality predictions in complex urban environments, especially under dynamic meteorological conditions.

FACE RECOGNITION WITH SIAMESE NEURAL NETWORKS

2025-02-21T06:13:47+00:00

The development of face recognition technologies has become increasingly critical due to the growing need for effective identification methods. Traditional techniques often struggle with variations in illumination, pose, and facial expressions, limiting their applicability in real-world scenarios. Recent advances in deep learning, particularly Convolutional Neural Networks (CNNs), have significantly improved performance on benchmark datasets. Siamese Neural Networks, a specialized class of CNNs, have emerged as a highly promising solution for face recognition, offering unparalleled capabilities in learning feature representations and similarity metrics. This study rigorously examines the efficiency of Siamese Neural Networks in face recognition across diverse datasets and real-time scenarios. Using three distinct face recognition datasets, the research evaluates the accuracy and robustness of the network under challenging conditions and assesses its ability to distinguish between similar and dissimilar faces in real-time applications. The results demonstrate the effectiveness of Siamese Neural Networks in handling variations in pose, illumination, and expressions, highlighting their potential to advance face recognition technology. These findings provide valuable insights into the practical applicability of Siamese Neural Networks in real-world contexts.

MODEL BASED SOLUTION FOR COMPUTING CHECKPOINTING INTERVAL FOR FAULT-TOLERANT ROLLBACK-RECOVERY IN ENTERPRISE SERVERS

2025-03-05T06:01:00+00:00

Recently, reliable information processing has become a relevant topic with the increase of digitalization. It is especially essential for enterprises that process huge amounts of data every day. These processes require stability and reliability as their interruption might lead to various security issues. In order to tackle this, there are fault-tolerance algorithms that are specifically designed to prevent or recover faults. This paper focuses on developing a heuristic solution to find optimal checkpointing interval for rollback-recovery fault-tolerance algorithm. Authors propose a heuristic solution that utilizes CPU capabilities to determine how often checkpointing should be taken for reliable information processing. Research provides statistics and predictions of major research organizations highlighting the relevance of the topic. Paper also reviews related work devoted to this area of research, providing comparisons and overall analysis. The results of the work show that the proposed calculation method introduces minimal performance overhead, averaging 0.04 seconds to the average service time, while maintaining fault tolerance of the process. Authors indicate that this solution is suitable for proof-of-concept systems to efficiently determine optimal interval for checkpointing.

INVESTIGATION OF EMERGENCY SITUATIONS IN ALMATY USING MACHINE LEARNING METHODS

2025-03-12T09:43:41+00:00

At present, the protection of the population from emergencies that occur daily and cause harm to people and the country's territory necessitates organizational measures for monitoring, research, forecasting, and prevention. This study focuses on different types of emergencies, including natural, social, and man-made disasters. With the increasing volume of information on the Internet, there is a growing need to analyze the continuous flow of data published on news websites. In this study, machine learning-based methods and approaches were utilized. A research analysis of emergency-related data was conducted, identifying the key factors influencing the frequency of incidents. Additionally, emergencies were classified and assessed based on their types. During the evaluation of various algorithms, the most effective machine learning methods were determined. Data was collected from open sources in text format and subsequently processed using natural language preprocessing techniques. By leveraging historical weather data for the city of Almaty, a correlation between emergencies and weather conditions was identified.

DEVELOPMENT OF HYBRID QUANTUM-CLASSICAL MODELS FOR COMPUTER VISION

2025-03-11T04:11:10+00:00

This research explores the integration of quantum computing with classical machine learning to enhance data classification tasks using Quantum Neural Networks (QNN) and Parameterized Quantum Circuits (PQC). The hybrid approach leverages the advantages of both quantum and classical systems to improve the efficiency and accuracy of data processing. In this model, data is encoded into qubits using amplitude encoding, representing input vectors as amplitudes of quantum states. The QNN is initialized by placing the qubits in superposition using Hadamard gates, followed by data encoding with parameterized rotational gates that map classical data to quantum states using rotation angles. PQC plays a central role by applying layers of parameterized quantum operations to process data in the quantum space. These parameters are optimized during the training process, where a quadratic loss function minimizes the error between the predicted quantum states and the true class labels using gradient descent. Experiments conducted on the MNIST dataset show that the hybrid the hybrid quantum-classical neural network (QCNN) with PQC achieves a classification accuracy of over 95%, highlighting its potential in machine learning applications. The results demonstrate that integrating quantum computing with classical machine learning enhances performance in complex data analysis tasks due to the exponential growth of quantum state space and the parallelism of quantum systems, making hybrid models promising for computer vision and classification tasks.

DEVELOPMENT OF A DEEP LEARNING MODEL FOR FORECASTING AND OPTIMIZING RIDE-SHARING ROUTES

2025-03-11T10:47:23+00:00

The study investigates the potential use of machine learning (ML) technologies, including Recurrent Neural Networks (RNNs), in ride-sharing and urban mobility optimization. Advanced deep learning (DL) models can solve growing challenges in urban areas, such as road safety, environmental pollution, and traffic congestion. Three different RNN architectures (SimpleRNN, LSTM, GRU) are compared to predict trips with their pickup and drop-off points. According to the assessment metrics, GRU shows better results in terms of Mean Haversine Distance (6.450 km) than SimpleRNN (7.156 km) and LSTM (6.569 km). Moreover, the GRU model surpasses other models in other indicators, such as MSE (0.0010) and MAE (0.0211). In addition, OSRM API is used to build routes between predicted pickup and drop-off points, as well as to optimize ride-sharing routes using real-time geographic data. The study highlights that ML approaches, in particular DL, can be used to solve problems related to urban mobility by improving transport efficiency and reducing traffic. The study results provide recommendations for developing urban transport systems using data-driven approaches to enhance ride-sharing opportunities.