Improved RPCA Method via Fractional Function-Based Structure and Its Application

Information, Journal Year: 2025, Volume and Issue: 16(1), P. 69 - 69

Published: Jan. 20, 2025

With the advancement of oil logging techniques, vast amounts data have been generated. However, this often contains significant redundancy and noise. The must be denoised before it is used for recognition. Hence, paper proposed an improved robust principal component analysis algorithm (IRPCA) denoising, which addresses problems various noises in acquisition limitations conventional processing methods. IRPCA enhances both efficiency model accuracy low-rank matrix recovery. This improvement achieved primarily by introducing approximate zero norm based on fractional function structure adding weighted kernel parametrization penalty terms to enhance model’s capability handling complex matrices. efficacy has verified through simulation experiments, demonstrating its superiority over widely RPCA algorithm. We then present a denoising method tailored characteristics first involves segregation original acquire background foreground information. information subsequently further separated isolate factual noise, resulting data. results indicate that practical effective when applied actual

Language: Английский

IT-RUDA: Information Theory Assisted Robust Unsupervised Domain Adaptation

ACM Transactions on Intelligent Systems and Technology, Journal Year: 2025, Volume and Issue: unknown

Published: April 28, 2025

Domain adaptation is a well-studied field in machine learning. Distribution shift between train (source) and test (target) datasets common problem encountered learning applications. One approach to resolve this issue use the Unsupervised Adaptation (UDA) technique that carries out knowledge transfer from label-rich source domain an unlabeled target domain. Outliers exist either or can introduce additional challenges when using UDA practice. In paper, \(\alpha\) -divergence used as measure minimize discrepancy distributions while inheriting robustness, adjustable with single parameter , prominent feature of measure. Here, it shown other well-known divergence-based techniques be derived special cases proposed method. Furthermore, theoretical upper bound for loss terms joint two domains. The robustness method validated through testing on several benchmarked open-set partial setups where extra classes existing are considered outliers. Code publicly available at https://github.com/rashidis/IT-RUDA .

Language: Английский

Evaluating efficiency in water and sewerage services: An integrated DEA approach with DOE and PCA

The Science of The Total Environment, Journal Year: 2024, Volume and Issue: 959, P. 178288 - 178288

Published: Dec. 31, 2024

Evaluating the performance of service organizations like Water and Sewerage companies is essential for optimal operations, high-quality service, cost efficiency. This paper introduces a model using data envelopment analysis (DEA) to assess efficiency operational units within such companies. The selection key indicators complicated by numerous inputs outputs, each affecting systems activities differently. To enhance DEA due imbalance between number inputs/outputs under evaluation, this research integrates design experiments (DOE) principal component (PCA) variable screening reduction, creating new linear combinations with minimal information loss. These methods represent direction in handling variables models. Addressing unit heterogeneity removing environmental factors from reduces errors. A case study showed that some can achieve high fewer more valuable outputs. findings offered managerial insights informed decision-making strategic planning, optimizing resources line company's mission vision. methodology ultimately improves reliability, customer satisfaction, sustainability. graphical abstract has been simplified readability focus on primary methodological advances. It emphasizes integration PCA dimensionality DOE scereening, evaluation.

Language: Английский