PeerJ Computer Science最新文献

Graphic design, as a product of the burgeoning new media era, has seen its users' requirements for images continuously evolve. However, external factors such as light and noise often cause graphic design images to become distorted during acquisition. To enhance the definition of these images, this paper introduces a novel image enhancement model based on visual features. Initially, a histogram equalization (HE) algorithm is applied to enhance the graphic design images. Subsequently, image feature extraction is performed using a dual-flow network comprising convolutional neural network (CNN) and Transformer architectures. The CNN employs a residual dense block (RDB) to embed spatial local structure information with varying receptive fields. An improved attention mechanism module, attention feature fusion (AFF), is then introduced to integrate the image features extracted from the dual-flow network. Finally, through image perception quality guided adversarial learning, the model adjusts the initial enhanced image's color and recovers more details. Experimental results demonstrate that the proposed algorithm model achieves enhancement effects exceeding 90% on two large image datasets, which represents a 5%-10% improvement over other models. Furthermore, the algorithm exhibits superior performance in terms of peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) image quality evaluation metrics. Our findings indicate that the fusion model significantly enhances image quality, thereby advancing the field of graphic design and showcasing its potential in cultural and creative product design.

Anomalies are the existential abnormalities in data, the identification of which is known as anomaly detection. The absence of timely detection of anomalies may affect the key processes of decision-making, fraud detection, and automated classification. Most of the existing models of anomaly detection utilize the traditional way of tokenizing and are computationally costlier, mainly if the outliers are to be extracted from a large script. This research work intends to propose an unsupervised, all-MiniLM-L6-v2-based system for the detection of outliers. The method makes use of centroid embeddings to extract outliers in high-variety, large-volume data. To avoid mistakenly treating novelty as an outlier, the Minimum Covariance Determinant (MCD) based approach is followed to count the novelty of the input script. The proposed method is implemented in a Python project, App. for Anomalies Detection (AAD). The system is evaluated by two non-related datasets-the 20 newsgroups text dataset and the SMS spam collection dataset. The robust accuracy (94%) and F1 score (0.95) revealed that the proposed method could effectively trace anomalies in a comparatively large script. The process is applicable in extracting meanings from textual data, particularly in the domains of human resource management and security.

Chronic kidney disease (CKD) involves numerous variables, but only a few significantly impact the classification task. The statistically equivalent signature (SES) method, inspired by constraint-based learning of Bayesian networks, is employed to identify essential features in CKD. Unlike conventional feature selection methods, which typically focus on a single set of features with the highest predictive potential, the SES method can identify multiple predictive feature subsets with similar performance. However, most feature selection (FS) classifiers perform suboptimally with strongly correlated data. The FS approach faces challenges in identifying crucial features and selecting the most effective classifier, particularly in high-dimensional data. This study proposes using the Least Absolute Shrinkage and Selection Operator (LASSO) in conjunction with the SES method for feature selection in CKD identification. Following this, an ensemble deep-learning model combining long short-term memory (LSTM) and gated recurrent unit (GRU) networks is proposed for CKD classification. The features selected by the hybrid feature selection method are fed into the ensemble deep-learning model. The model's performance is evaluated using accuracy, precision, recall, and F1 score metrics. The experimental results are compared with individual classifiers, including decision tree (DT), Random Forest (RF), logistic regression (LR), and support vector machine (SVM). The findings indicate a 2% improvement in classification accuracy when using the proposed hybrid feature selection method combined with the LSTM and GRU ensemble deep-learning model. Further analysis reveals that certain features, such as HEMO, POT, bacteria, and coronary artery disease, contribute minimally to the classification task. Future research could explore additional feature selection methods, including dynamic feature selection that adapts to evolving datasets and incorporates clinical knowledge to enhance CKD classification accuracy further.

Plant diseases threaten agricultural sustainability by reducing crop yields. Rapid and accurate disease identification is crucial for effective management. Recent advancements in artificial intelligence (AI) have facilitated the development of automated systems for disease detection. This study focuses on enhancing the classification of diseases and estimating their severity in coffee leaf images. To do so, we propose a novel approach as the preprocessing step for the classification in which enhanced multivariance product representation (EMPR) is used to decompose the considered image into components, a new image is constructed using some of those components, and the contrast of the new image is enhanced by applying high-dimensional model representation (HDMR) to highlight the diseased parts of the leaves. Popular convolutional neural network (CNN) architectures, including AlexNet, VGG16, and ResNet50, are evaluated. Results show that VGG16 achieves the highest classification accuracy of approximately 96%, while all models perform well in predicting disease severity levels, with accuracies exceeding 85%. Notably, the ResNet50 model achieves accuracy levels surpassing 90%. This research contributes to the advancement of automated crop health management systems.

Federated learning (FL) is a popular method where edge devices work together to train machine learning models. This study introduces an efficient network for analyzing healthcare records. It uses VPN technology and applies a federated learning approach over a wireless backhaul network. The study compares different wireless backhaul channels, including terahertz (THz), E/V band (mmWave), and microwave, for their effectiveness. We looked closely at a suggested FL network that uses VPN technology over awireless backhaul network. We compared it with the standard method and found that using the FedAvg algorithm with Terahertz (THz) for communication gave the best accuracy. The time it took to reach a conclusion improved a lot, going from 55 seconds to an impressive 38 seconds. This emphasizes how having a faster communication link makes FL networks work much better. Furthermore, a three-step plan was executed to boost security, adopting a multi-layered method to safeguard the FL network and its confidential data. The first step involves integrating a private network into the current telecom infrastructure, establishing an initial layer of security. To enhance security further, licensed frequency channels are introduced, providing an extra layer of protection. The highest level of security is achieved by combining a private network with licensed frequency channels, complemented by an additional layer of security through VPN-based measures. This comprehensive strategy ensures the application of strong security protocols.

The rise of the Internet of Things (IoT) and Industry 2.0 has spurred a growing need for extensive data computing, and Spark emerged as a promising Big Data platform, attributed to its distributed in-memory computing capabilities. However, practical heavy workloads often lead to memory bottleneck issues in the Spark platform. This results in resilient distributed datasets (RDD) eviction and, in extreme cases, violent memory contentions, causing a significant degradation in Spark computational efficiency. To tackle this issue, we propose an adaptive memory reservation (AMR) strategy in this article, specifically designed for heavy workloads in the Spark environment. Specifically, we model optimal task parallelism by minimizing the disparity between the number of tasks completed without blocking and the number completed in regular rounds. Optimal memory for task parallelism is determined to establish an efficient execution memory space for computational parallelism. Subsequently, through adaptive execution memory reservation and dynamic adjustments, such as compression or expansion based on task progress, the strategy ensures dynamic task parallelism in the Spark parallel computing process. Considering the cost of RDD cache location and real-time memory space usage, we select suitable storage locations for different RDD types to alleviate execution memory pressure. Finally, we conduct extensive laboratory experiments to validate the effectiveness of AMR. Results indicate that, compared to existing memory management solutions, AMR reduces the execution time by approximately 46.8%.

Background: Multiple-choice questions (MCQs) are one of the most used assessment formats. However, creating MCQs is a challenging task, particularly when formulating the distractor. Numerous studies have proposed automatic distractor generation. However, there has been no literature review to summarize and present the current state of research in this field. This study aims to perform a systematic literature review to identify trends and the state of the art of automatic distractor generation studies.

Methodology: We conducted a systematic literature following the Kitchenham framework. The relevant literature was retrieved from the ACM Digital Library, IEEE Xplore, Science Direct, and Scopus databases.

Results: A total of 60 relevant studies from 2009 to 2024 were identified and extracted to answer three research questions regarding the data sources, methods, types of questions, evaluation, languages, and domains used in the automatic distractor generation research. The results of the study indicated that automatic distractor generation has been growing with improvement and expansion in many aspects. Furthermore, trends and the state of the art in this topic were observed.

Conclusions: Nevertheless, we identified potential research gaps, including the need to explore further data sources, methods, languages, and domains. This study can serve as a reference for future studies proposing research within the field of automatic distractor generation.

The creation of 3D animation increasingly prioritizes the enhancement of character effects, narrative depth, and audience engagement to address the growing demands for visual stimulation, cultural enrichment, and interactive experiences. The advancement of virtual reality (VR) animation is anticipated to require sustained collaboration among researchers, animation experts, and hardware developers over an extended period to achieve full maturity. This article explores the use of Virtual Reality Modeling Language (VRML) in generating 3D stereoscopic forms and environments, applying texture mapping, optimizing lighting effects, and establishing interactive user responses, thereby enriching the 3D animation experience. VRML's functionality is further expanded through the integration of script programs in languages such as Java, JavaScript, and VRML Script via the Script node. The implementation of fuzzy model recognition within 3D animation simulations enhances the identification of textual, musical, and linguistic elements, resulting in improved frame rates. This study also analyzes the real-time correlation between the number of polygons and frame rates in a virtual museum animation scene. The findings demonstrate that the frame rate of the 3D animation within this virtual setting consistently exceeds 40 frames per second, thereby ensuring robust real-time performance, preserving the quality of 3D models, and optimizing rendering speed and visual effects without affecting the system's responsiveness to additional functions.

In recent years, deep learning models have become predominant methods for computer vision tasks, but the large computation and storage requirements of many models make them challenging to deploy on devices with limited resources. Knowledge distillation (KD) is a widely used approach for model compression. However, when applied in the object detection problems, the existing KD methods either directly applies the feature map or simply separate the foreground from the background by using a binary mask, aligning the attention between the teacher and the student models. Unfortunately, these methods either completely overlook or fail to thoroughly eliminate noise, resulting in unsatisfactory model accuracy for student models. To address this issue, we propose a foreground separation distillation (FSD) method in this paper. The FSD method enables student models to distinguish between foreground and background using Gaussian heatmaps, reducing irrelevant information in the learning process. Additionally, FSD also extracts the channel feature by converting the spatial feature maps into probabilistic forms to fully utilize the knowledge in each channel of a well-trained teacher. Experimental results demonstrate that the YOLOX detector enhanced with our distillation method achieved superior performance on both the fall detection and the VOC2007 datasets. For example, YOLOX with FSD achieved 73.1% mean average precision (mAP) on the Fall Detection dataset, which is 1.6% higher than the baseline. The code of FSD is accessible via https://doi.org/10.5281/zenodo.13829676.

In the context of high-quality economic development, technological innovation has emerged as a fundamental driver of socio-economic progress. The consequent proliferation of science and technology news, which acts as a vital medium for disseminating technological advancements and policy changes, has attracted considerable attention from technology management agencies and innovation organizations. Nevertheless, online science and technology news has historically exhibited characteristics such as limited scale, disorderliness, and multi-dimensionality, which is extremely inconvenient for users of deep application. While single-label classification techniques can effectively categorize textual information, they face challenges in leading science and technology news classification due to a lack of a hierarchical knowledge framework and insufficient capacity to reveal knowledge integration features. This study proposes a hierarchical multi-label classification model for science and technology news, enhanced by heterogeneous graph semantics. The model captures multi-dimensional themes and hierarchical structural features within science and technology news through a hierarchical transmission module. It integrates graph convolutional networks to extract node information and hierarchical relationships from heterogeneous graphs, while also incorporating prior knowledge from domain knowledge graphs to address data scarcity. This approach enhances the understanding and classification capabilities of the semantics of science and technology news. Experimental results demonstrate that the model achieves precision, recall, and F1 scores of 84.21%, 88.89%, and 86.49%, respectively, significantly surpassing baseline models. This research presents an innovative solution for hierarchical multi-label classification tasks, demonstrating significant application potential in addressing data scarcity and complex thematic classification challenges.