PeerJ Computer Science最新文献

As the aesthetic appreciation for art continues to grow, there is an increased demand for precision and detailed control in sculptural works. The advent of 3D laser scanning technology introduces transformative new tools and methodologies for refining correction systems in sculpture design. This article proposes a feature point matching algorithm based on fragment measurement and the iterative closest point (ICP) methodology, leveraging 3D laser scanning technology, namely Fragment Measurement Iterative Closest Point Feature Point Matching (FM-ICP-FPM). The FM-ICP-FPM approach uses the overlapping area of the two sculpture perspectives as a reference for attaching feature points. It employs the 3D measurement system to capture physical point cloud data from the two surfaces to enable the initial alignment of feature points. Feature vectors are generated by segmenting the region around the feature points and computing the intra-block gradient histogram. Subsequently, distance threshold conditions are set based on the constructed feature vectors and the preliminary feature point matches established during the coarse alignment to achieve precise feature point matching. Experimental results demonstrate the exceptional performance of the FM-ICP-FPM algorithm, achieving a sampling interval of 200. The correct matching rate reaches an impressive 100%, while the mean translation error (MTE) is a mere 154 mm, and the mean rotation angle error (MRAE) is 0.065 degrees. The indicator represents the degree of deviation in translation and rotation of the registered model, respectively. These low error values demonstrate that the FM-ICP-FPM algorithm excels in registration accuracy and can generate highly consistent three-dimensional models.

Computational drug-target affinity prediction is important for drug screening and discovery. Currently, self-supervised learning methods face two major challenges in drug-target affinity prediction. The first difficulty lies in the phenomenon of sample mismatch: self-supervised learning processes drug and target samples independently, while actual prediction requires the integration of drug-target pairs. Another challenge is the mismatch between the broadness of self-supervised learning objectives and the precision of biological mechanisms of drug-target affinity (i.e., the induced-fit principle). The former focuses on global feature extraction, while the latter emphasizes the importance of local precise matching. To address these issues, an adaptive self-supervised learning-based drug-target affinity prediction (ASSLDTA) was designed. ASSLDTA integrates a novel adaptive self-supervised learning (ASSL) module with a high-level feature learning network to extract the feature. The ASSL leverages a large amount of unlabeled training data to effectively capture low-level features of drugs and targets. Its goal is to maximize the retention of original feature information, thereby bridging the objective gap between self-supervised learning and drug-target affinity prediction and alleviating the sample mismatch problem. The high-level feature learning network, on the other hand, focuses on extracting effective high-level features for affinity prediction through a small amount of labeled data. Through this two-stage feature extraction design, each stage undertakes specific tasks, fully leveraging the advantages of each model while efficiently integrating information from different data sources, providing a more accurate and comprehensive solution for drug-target affinity prediction. In our experiments, ASSLDTA is much better than other deep methods, and the result of ASSLDTA is significantly increased by learning adaptive self-supervised learning-based features, which validates the effectiveness of our ASSLDTA.

Recent studies have shown that combining Transformer and conditional strategies to deal with offline reinforcement learning can bring better results. However, in a conventional reinforcement learning scenario, the agent can receive a single frame of observations one by one according to its natural chronological sequence, but in Transformer, a series of observations are received at each step. Individual features cannot be extracted efficiently to make more accurate decisions, and it is still difficult to generalize effectively for data outside the distribution. We focus on the characteristic of few-shot learning in pre-trained models, and combine prompt learning to enhance the ability of real-time policy adjustment. By sampling the specific information in the offline dataset as trajectory samples, the task information is encoded to help the pre-trained model quickly understand the task characteristics and the sequence generation paradigm to quickly adapt to the downstream tasks. In order to understand the dependencies in the sequence more accurately, we also divide the fixed-size state information blocks in the input trajectory, extract the features of the segmented sub-blocks respectively, and finally encode the whole sequence into the GPT model to generate decisions more accurately. Experiments show that the proposed method achieves better performance than the baseline method in related tasks, can be generalized to new environments and tasks better, and effectively improves the stability and accuracy of agent decision making.

Background: As modern agricultural technology advances, the automated detection, classification, and harvesting of strawberries have become an inevitable trend. Among these tasks, the classification of strawberries stands as a pivotal juncture. Nevertheless, existing object detection methods struggle with substantial computational demands, high resource utilization, and reduced detection efficiency. These challenges make deployment on edge devices difficult and lead to suboptimal user experiences.

Methods: In this study, we have developed a lightweight model capable of real-time detection and classification of strawberry fruit, named the Strawberry Lightweight Feature Classify Network (SLFCNet). This innovative system incorporates a lightweight encoder and a self-designed feature extraction module called the Combined Convolutional Concatenation and Sequential Convolutional (C3SC). While maintaining model compactness, this architecture significantly enhances its feature decoding capabilities. To evaluate the model's generalization potential, we utilized a high-resolution strawberry dataset collected directly from the fields. By employing image augmentation techniques, we conducted experimental comparisons between manually counted data and the model's inference-based detection and classification results.

Results: The SLFCNet model achieves an average precision of 98.9% in the mAP@0.5 metric, with a precision rate of 94.7% and a recall rate of 93.2%. Notably, SLFCNet features a streamlined design, resulting in a compact model size of only 3.57 MB. On an economical GTX 1080 Ti GPU, the processing time per image is a mere 4.1 ms. This indicates that the model can smoothly run on edge devices, ensuring real-time performance. Thus, it emerges as a novel solution for the automation and management of strawberry harvesting, providing real-time performance and presenting a new solution for the automatic management of strawberry picking.

Warfarin, a commonly prescribed anticoagulant, poses significant dosing challenges due to its narrow therapeutic range and high variability in patient responses. This study applies advanced machine learning techniques to improve the accuracy of international normalized ratio (INR) predictions using the MIMIC-III dataset, addressing the critical issue of missing data. By leveraging dimensionality reduction methods such as principal component analysis (PCA) and t-distributed stochastic neighbor embedding (t-SNE), and advanced imputation techniques including denoising autoencoders (DAE) and generative adversarial networks (GAN), we achieved significant improvements in predictive accuracy. The integration of these methods substantially reduced prediction errors compared to traditional approaches. This research demonstrates the potential of machine learning (ML) models to provide more personalized and precise dosing strategies that reduce the risks of adverse drug events. Our method could integrate into clinical workflows to enhance anticoagulation therapy in cases of missing data, with potential applications in other complex medical treatments.

Machine-to-machine (M2M) communication within the Internet of Things (IoT) faces increasing security and efficiency challenges as networks proliferate. Existing approaches often struggle with balancing robust security measures and energy efficiency, leading to vulnerabilities and reduced performance in resource-constrained environments. To address these limitations, we propose SAFE-CAST, a novel secure AI-federated enumeration for clustering-based automated surveillance and trust framework. This study addresses critical security and efficiency challenges in M2M communication within the context of IoT. SAFE-CAST integrates several innovative components: (1) a federated learning approach using Lloyd's K-means algorithm for secure clustering, (2) a quality diversity optimization algorithm (QDOA) for secure channel selection, (3) a dynamic trust management system utilizing blockchain technology, and (4) an adaptive multi-agent reinforcement learning for context-aware transmission scheme (AMARLCAT) to minimize latency and improve scalability. Theoretical analysis and extensive simulations using network simulator (NS)-3.26 demonstrate the superiority of SAFE-CAST over existing methods. The results show significant improvements in energy efficiency (21.6% reduction), throughput (14.5% increase), security strength (15.3% enhancement), latency (33.9% decrease), and packet loss rate (12.9% reduction) compared to state-of-the-art approaches. This comprehensive solution addresses the pressing need for robust, efficient, and secure M2M communication in the evolving landscape of IoT and edge computing.

The global spread of SARS-CoV-2 has prompted a crucial need for accurate medical diagnosis, particularly in the respiratory system. Current diagnostic methods heavily rely on imaging techniques like CT scans and X-rays, but identifying SARS-CoV-2 in these images proves to be challenging and time-consuming. In this context, artificial intelligence (AI) models, specifically deep learning (DL) networks, emerge as a promising solution in medical image analysis. This article provides a meticulous and comprehensive review of imaging-based SARS-CoV-2 diagnosis using deep learning techniques up to May 2024. This article starts with an overview of imaging-based SARS-CoV-2 diagnosis, covering the basic steps of deep learning-based SARS-CoV-2 diagnosis, SARS-CoV-2 data sources, data pre-processing methods, the taxonomy of deep learning techniques, findings, research gaps and performance evaluation. We also focus on addressing current privacy issues, limitations, and challenges in the realm of SARS-CoV-2 diagnosis. According to the taxonomy, each deep learning model is discussed, encompassing its core functionality and a critical assessment of its suitability for imaging-based SARS-CoV-2 detection. A comparative analysis is included by summarizing all relevant studies to provide an overall visualization. Considering the challenges of identifying the best deep-learning model for imaging-based SARS-CoV-2 detection, the article conducts an experiment with twelve contemporary deep-learning techniques. The experimental result shows that the MobileNetV3 model outperforms other deep learning models with an accuracy of 98.11%. Finally, the article elaborates on the current challenges in deep learning-based SARS-CoV-2 diagnosis and explores potential future directions and methodological recommendations for research and advancement.

The OptiFlow framework introduces a novel approach for enhancing usability evaluation and optimization known as OptiFlow. This framework combines heuristic evaluation with a web-based platform to provide a comprehensive method for assessing and optimizing user experiences in online applications. The architecture of OptiFlow incorporates key components, including the user, website, web service, and library, enabling seamless interaction and data exchange. A set of 240 usability guidelines, derived from a multidisciplinary expert collaboration, are systematically categorized into 15 usability categories, aligned with established design principles. Guidelines within OptiFlow are assigned implementation levels: "Green" for easily implementable guidelines, "Amber" for moderately complex ones, and "Red" for highly complex guidelines. These levels prioritize tasks based on complexity and feasibility. The framework's integration of guidelines into a structured SQL database simplifies implementation challenges, and the "execute" function systematically assesses website adherence to guidelines, resulting in True, False, or Null outcomes. Usability assessment outcomes are presented through categorized and prioritized data views for each implementation level, allowing stakeholders to address high-priority concerns efficiently. The OptiFlow framework represents an innovative approach to usability evaluation, fostering enriched user experiences and finely tuned digital interfaces. Future advancements may include additional rule types and the integration of advanced technologies for tackling intricate usability challenges. Ultimately, OptiFlow paves the way for proactive user experience enhancement and digital interface optimization in an ever-evolving digital landscape.

Choreographies provide a clear way to specify the intended communication behaviour of concurrent and distributed systems. Previous theoretical work investigated the translation of choreographies into (models of) programs based on message passing. However, existing theories still present a gap between how they model communications-using channel names à la CCS or $\pi$ -calculus-and implementations-which use lower-level mechanisms for message routing. We start bridging this gap with a new formal framework called Applied Choreographies. In Applied Choreographies, developers write choreographies in a familiar syntax (from previous work) and reason about their behaviour through simple, abstract name-based communication semantics. The framework offers state-of-the-art features of choreographic models, e.g., modular programming supported via choreographic types. To provide its correctness guarantee, Applied Choreographies comes with a compilation procedure that transforms a choreography into a low-level, implementation-adherent calculus of Service-Oriented Computing (SOC). To manage the complexity of the compilation, we divide its formalisation and proof into three stages, respectively dealing with: (a) the translation of name-based communications into their SOC equivalents, namely, using correlation mechanisms based on message data; (b) the projection of the given choreography into a composition of partial, single-participant choreographies (towards their translation into SOC processes); (c) the translation of partial choreographies and the distribution of global, choreography-level state into local SOC processes. We provide behavioural correspondence results for each stage. Thus, given a choreography specification, we guarantee to synthesise its faithful service-oriented implementation.

These days, location-based services, or LBS, are used for various consumer applications, including indoor localization. Due to the ease with which Wi-Fi can be accessed in various interior settings, there has been increasing interest in Wi-Fi-based indoor localisation. Deep learning in indoor localisation systems that use channel state information (CSI) fingerprinting has seen widespread adoption. Usually, these systems comprise two primary components: a positioning network and a tracking system. The positioning network is responsible for learning the planning from high-dimensional CSI to physical positions, and the following system uses historical CSI to decrease positioning error. This work presents a novel localization method that combines high accuracy and generalizability. However, existing convolutional neural network (CNN) fingerprinting placement algorithms have a limited receptive area, limiting their effectiveness since important data in CSI has not been thoroughly explored. We offer a unique attention-augmented residual CNN to remedy this issue so that the data acquired and the global context in CSI may be utilized to their full potential. On the other hand, while considering the generalizability of a monitoring device, we uncouple the scheme from the CSI environments to make it feasible to use a single tracking system across all contexts. To be more specific, we recast the tracking issue as a denoising task and then used a deep route before solving it. The findings illuminate perspectives and realistic interpretations of the residual attention-based CNN (RACNN) in device-free Wi-Fi indoor localization using channel state information (CSI) fingerprinting. In addition, we study how the precision change of different inertial dimension units may negatively influence the tracking performance, and we implement a solution to the problem of exactness variance. The proposed RACNN model achieved a localization accuracy of 99.9%, which represents a significant improvement over traditional methods such as K-nearest neighbors (KNN) and Bayesian inference. Specifically, the RACNN model reduced the average localization error to 0.35 m, outperforming these traditional methods by approximately 14% to 15% in accuracy. This improvement demonstrates the model's ability to handle complex indoor environments and proves its practical applicability in real-world scenarios.