Lesia Mochurad

Microgrids with a high share of renewable generation require operational energy management strategies that can continuously balance consumption, local generation, and market interaction while maint...

This paper proposes and experimentally evaluates a multi-level CPU-oriented execution method for high-throughput portion-based processing of file-backed Variant Call Format (VCF) data and automated mutation classification. The approach is based on a formally defined local processing scheme and integrates three coordinated levels of parallelism: block-based partitioning of file-backed VCF portions read sequentially into localized fragments with data-level parallel processing; task-level decomposition of feature construction into independent transformations; and execution-level specialization via JIT compilation of numerical kernels. To prevent performance degradation caused by nested parallelism, a resource-control mechanism is introduced as an execution rule that bounds effective parallelism and mitigates oversubscription, improving throughput stability on a single multi-core CPU node. Experiments on a public chromosome-17 VCF dataset for BRCA1-region pathogenicity classification demonstrate that the proposed multi-level local CPU execution (parsing/filtering, feature construction, and JIT-specialized numeric kernels) reduces runtime from 291.25 s (sequential) to 73.82 s, yielding a 3.95× speedup. When combined with resource-coordinated parallel model training, the end-to-end runtime further decreases to 51.18 s, corresponding to a 5.69× speedup, while preserving classification quality (accuracy 0.8483, precision 0.8758, recall 0.8261, F1 0.8502). A stage-wise ablation analysis quantifies the contribution of each execution level and confirms consistent scaling under resource-bounded execution.

The rapid growth of high-dimensional biometric data requires fast and accurate similarity search methods for real-time applications. This study proposes, for the first time, two efficient parallel implementations of the exact Sorting-based Nearest Neighbor (SNN) algorithm using OpenMP for CPUs and CUDA for GPUs. Comparative evaluation against conventional exact search methods—k-d tree and ball tree—on LFW embeddings, including FaceNet512 and VGG-Face, demonstrates an up to 58× speedup on GPUs while maintaining full accuracy. Analysis of the full recognition pipeline shows that parallelization reduces search times to about 27% of total processing, highlighting the method’s stability and efficiency for modern embeddings. These results confirm the applicability of the proposed approaches for real-time biometric identification, with potential extensions to streaming data, hybrid computing environments, and other high-dimensional representations.

In the context of the growing volume and complexity of data, traditional methods of energy consumption forecasting, such as Recurrent Neural Networks (RNN), face computational complexity issues that limit their real-time application. This also complicates the effective management of energy systems. In this work, a new model is proposed that combines the advantages of Temporal Convolutional Networks (TCN) and Quasi-Recurrent Neural Networks (QRNN) for energy consumption forecasting. TCN allows for effective processing of long time series, capturing essential temporal dependencies. Meanwhile, QRNN reduces computational costs through parallelization of operations and an optimized architecture. The effectiveness of the proposed model has been assessed in comparison with traditional methods such as Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks, as well as other convolutional approaches. Experimental results show that the proposed TCN-QRNN model outperforms traditional methods by 40% in accuracy compared to LSTM and by 8% in terms of metrics such as Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) compared to TCN-LSTM, while reducing data processing time by 30%. Additionally, the model has a significantly smaller number of parameters than LSTM and GRU, making it suitable for environments with limited computational resources. The proposed model ensures a high level of energy consumption forecasting accuracy while significantly reducing processing time, making it promising for use in real-world energy systems.

Many people admire the Japanese language and culture, but mastering the language’s writing system, particularly handwritten kanji, presents a significant challenge. Furthermore, translating historical manuscripts containing archaic or rare kanji requires specialized expertise. To address this, we designed a new model for handwritten kanji recognition based on the concept of cross-language transfer learning using a Preact ResNet-18 architecture. The model was pretrained in a Chinese dataset and subsequently fine-tuned in a Japanese dataset. We also adapted and evaluated two fine-tuning strategies: unfreezing only the last layer and unfreezing all the layers during fine-tuning. During the implementation of our training algorithms, we trained a model with the CASIA-HWDB dataset with handwritten Chinese characters and used its weights to initialize models that were fine-tuned with a Kuzushiji-Kanji dataset that consists of Japanese handwritten kanji. We investigated the effectiveness of the developed model when solving a multiclass classification task for three subsets with the one hundred fifty, two hundred, and three hundred most-sampled classes and showed an improvement in the recognition accuracy and an enhancement in a number of recognizable kanji with the proposed model compared to those of the existing methods. Our best model achieved 97.94% accuracy for 150 kanji, exceeding the previous SOTA result by 1.51%, while our best model for 300 kanji achieved 97.62% accuracy (exceeding the 150-kanji SOTA accuracy by 1.19% while doubling the class count). This confirms the effectiveness of our proposed model and establishes new benchmarks in handwritten kanji recognition, both in terms of accuracy and the number of recognizable kanji.

Introduction: Accurate segmentation of anatomical structures in chest X-ray images remains challenging, especially for regions with low contrast and overlapping structures. This limitation significantly affects the diagnosis of cardiothoracic diseases. Existing deep learning methods often struggle with preserving structural boundaries, leading to segmentation artifacts.Methods: To address these challenges, I propose a novel segmentation approach that integrates contour detection techniques with the U-net deep learning architecture. Specifically, the method employs Sobel and Scharr edge detection filters to enhance structural boundaries in chest X-ray images before segmentation. The pipeline involves pre-processing using contour detection, followed by segmentation with a U-net model trained to identify lungs, heart, and clavicles.Results: Experimental evaluation demonstrated that using edge-enhancing filters, particularly the Sobel operator, leads to a marked improvement in segmentation accuracy. For lung segmentation, the model achieved an accuracy of 99.26%, a Dice coefficient of 98.88%, and a Jaccard index of 97.54%. Heart segmentation results included 99.47% accuracy and 94.14% Jaccard index, while clavicle segmentation reached 99.79% accuracy and 89.57% Jaccard index. These results consistently outperform the baseline U-net model without edge enhancement.Discussion: The integration of contour detection methods with the U-net model significantly improves the segmentation quality of complex anatomical regions in chest X-rays. Among the tested filters, the Sobel operator proved to be the most effective in enhancing boundary information and reducing segmentation artifacts. This approach offers a promising direction for more accurate and robust computer-aided diagnosis systems in radiology.