In recent years in Taiwan, scholars who study financial bankruptcy have mostly focused on individual listed and over-the-counter (OTC) industries or the entire industry, while few have studied the independent electronics industry. Thus, this study investigated the application of an advanced hybrid Z-score bankruptcy prediction model in selecting financial ratios of listed companies in eight related electronics industries (semiconductor, computer, and peripherals, photoelectric, communication network, electronic components, electronic channel, information service, and other electronics industries) using data from 2000 to 2019. Based on 22 financial ratios of condition attributes and one decision attribute recommended and selected by experts and in the literature, this study used five classifiers for binary logistic regression analysis and in the decision tree. The experimental results show that for the Z-score model, samples analyzed using the five classifiers in five groups (1:1–5:1) of different ratios of companies, the bagging classifier scores are worse (40.82%) than when no feature selection method is used, while the logistic regression classifier and decision tree classifier (J48) result in better scores. However, it is significant that the bagging classifier score improved to over 90% after using the feature selection technique. In conclusion, it was found that the feature selection method can be effectively applied to improve the prediction accuracy, and three financial ratios (the liquidity ratio, debt ratio, and fixed assets turnover ratio) are identified as being the most important determinants affecting the prediction of financial bankruptcy in providing a useful reference for interested parties to evaluate capital allocation to avoid high investment risks.

With the widespread application of wearable sensors in health monitoring and human–computer interaction, deep learning-based human activity recognition (HAR) research faces challenges such as the effective extraction of multi-scale temporal features and the enhancement of robustness against noise in multi-source data. This study proposes the TGA-HAR (TCN-GRU-Attention-HAR) model. The TGA-HAR model integrates Temporal Convolutional Neural Networks and Recurrent Neural Networks by constructing a hierarchical feature abstraction architecture through cascading Temporal Convolutional Network (TCN) and Bidirectional Gated Recurrent Unit (BiGRU) layers for complex activity recognition. This study utilizes TCN layers with dilated convolution kernels to extract multi-order temporal features. This study utilizes BiGRU layers to capture bidirectional temporal contextual correlation information. To further optimize feature representation, the TGA-HAR model introduces residual connections to enhance the stability of gradient propagation and employs an adaptive weighted attention mechanism to strengthen feature representation. The experimental results of this study demonstrate that the model achieved test accuracies of 99.37% on the WISDM dataset, 95.36% on the USC-HAD dataset, and 96.96% on the PAMAP2 dataset. Furthermore, we conducted tests on datasets collected in real-world scenarios. This method provides a highly robust solution for complex human activity recognition tasks.

In this study, a novel adaptive fuzzy weighted mean filter (AFWMF) model based on the directional median technique and fuzzy inference is presented for solving the restoring high-ratio random-valued noise in image processing. This study aims, not only to obtain information from each direction of the filtering window, but also to gain information from every pixel of the filtering windows completely. Thus, in order to implement preserving details and textures for better restoration in high-noise cases, this study utilizes the directional median to build the membership function in fuzzy inference dynamically, then calculates the weighted window corresponding to the filtering window using fuzzy inference to represent the importance of valuable pixels. Finally, the restoration pixel is calculated using the weighted window and the filtering window for the weighted mean. Subsequently, this new AFWMF model significantly improves performances in the measurement of the peak signal to noise ratio (PSNR) value for preserving detail and fixed image in noise density within the range of 20–70% for the five well-known experimental images. In extensive experiments, this study also shows the better performance of identifying the proposed peak signal-to-removal noise ratio (PSRNR) and evaluating psycho-visual tests than other listed filter methods. Furthermore, the proposed AFWMF model also has a better structural similarity index measure (SSIM) value of another indicator. Conclusively, two interesting and meaning findings are identified: (1) the proposed AFWMF model is generally the best model among the 10 listed filtering methods for image processing in terms of the measurement of two quantitative indicators for both the PSNR and SSIM values; (2) different impulse noise densities should be made for different filtering methods, and thus, this is an important and interesting issue when aiming to identify an appropriate filtering model from a variety of images for processing various noise densities.

With the rise of Industry 4.0, Augmented Reality (AR) has become pivotal for human-robot collaboration. However, most industrial AR systems still rely on pre-defined tracked images or markers, limiting adaptability in unmodeled or dynamic environments. This paper proposes a novel Interactive Semantic-Augmented Reality (ISAR) framework that synergizes Edge AI and Cloud Vision-Language Models (VLMs). To ensure real-time performance, we implement a Dual-Thread Asynchronous Architecture on the robotic edge, decoupling video streaming from AI inference. We introduce a Confidence-Based Triggering Mechanism, where a cloud-based VLM is invoked only when edge detection confidence falls below a specific threshold. Instead of traditional image cropping, we employ a Visual Prompting strategy—overlaying bounding boxes on full-frame images—to preserve spatial context for accurate VLM semantic analysis. Finally, the generated insights are anchored to the physical world via Screen-to-World Raycasting without fiducial markers. This framework realizes a semantic-aware 'Intelligent Agent' that enhances Human-in-the-Loop (HITL) decision-making in complex industrial settings.