Frontiers in Big Data最新文献

Low-rank tensor completion (LRTC), which aims to complete missing entries from tensors with partially observed terms by utilizing the low-rank structure of tensors, has been widely used in various real-world issues. The core tensor nuclear norm minimization (CTNM) method based on Tucker decomposition is one of common LRTC methods. However, the CTNM methods based on Tucker decomposition often have a large computing cost due to the fact that the general factor matrix solving technique involves multiple singular value decompositions (SVDs) in each loop. To address this problem, this article enhances the method and proposes an effective CTNM method based on thin QR decomposition (CTNM-QR) with lower computing complexity. The proposed method extends the CTNM by introducing tensor versions of the auxiliary variables instead of matrices, while using the thin QR decomposition to solve the factor matrix rather than the SVD, which can save the computational complexity and improve the tensor completion accuracy. In addition, the CTNM-QR method's convergence and complexity are analyzed further. Numerous experiments in synthetic data, real color images, and brain MRI data at different missing rates demonstrate that the proposed method not only outperforms in terms of completion accuracy and visualization, but also conducts more efficiently than most state-of-the-art LRTC methods.

The k-nearest neighbors (KNN) regression method, known for its nonparametric nature, is highly valued for its simplicity and its effectiveness in handling complex structured data, particularly in big data contexts. However, this method is susceptible to overfitting and fit discontinuity, which present significant challenges. This paper introduces the random kernel k-nearest neighbors (RK-KNN) regression as a novel approach that is well-suited for big data applications. It integrates kernel smoothing with bootstrap sampling to enhance prediction accuracy and the robustness of the model. This method aggregates multiple predictions using random sampling from the training dataset and selects subsets of input variables for kernel KNN (K-KNN). A comprehensive evaluation of RK-KNN on 15 diverse datasets, employing various kernel functions including Gaussian and Epanechnikov, demonstrates its superior performance. When compared to standard KNN and the random KNN (R-KNN) models, it significantly reduces the root mean square error (RMSE) and mean absolute error, as well as improving R-squared values. The RK-KNN variant that employs a specific kernel function yielding the lowest RMSE will be benchmarked against state-of-the-art methods, including support vector regression, artificial neural networks, and random forests.

With the increasing popularity of Graph Neural Networks (GNNs) for predictive tasks on graph structured data, research on their explainability is becoming more critical and achieving significant progress. Although many methods are proposed to explain the predictions of GNNs, their focus is mainly on "how to generate explanations." However, other important research questions like "whether the GNN explanations are inaccurate," "what if the explanations are inaccurate," and "how to adjust the model to generate more accurate explanations" have gained little attention. Our previous GNN Explanation Supervision (GNES) framework demonstrated effectiveness on improving the reasonability of the local explanation while still keep or even improve the backbone GNNs model performance. In many applications instead of per sample explanations, we need to find global explanations which are reasonable and faithful to the domain data. Simply learning to explain GNNs locally is not an optimal solution to a global understanding of the model. To improve the explainability power of the GNES framework, we propose the Global GNN Explanation Supervision (GGNES) technique which uses a basic trained GNN and a global extension of the loss function used in the GNES framework. This GNN creates local explanations which are fed to a Global Logic-based GNN Explainer, an existing technique that can learn the global Explanation in terms of a logic formula. These two frameworks are then trained iteratively to generate reasonable global explanations. Extensive experiments demonstrate the effectiveness of the proposed model on improving the global explanations while keeping the performance similar or even increase the model prediction power.

Tuberculosis (TB) is a chronic and pathogenic disease that leads to life-threatening situations like death. Many people have been affected by TB owing to inaccuracy, late diagnosis, and deficiency of treatment. The early detection of TB is important to protect people from the severity of the disease and its threatening consequences. Traditionally, different manual methods have been used for TB prediction, such as chest X-rays and CT scans. Nevertheless, these approaches are identified as time-consuming and ineffective for achieving optimal results. To resolve this problem, several researchers have focused on TB prediction. Conversely, it results in a lack of accuracy, overfitting of data, and speed. For improving TB prediction, the proposed research employs the Selection Focal Fusion (SFF) block in the You Look Only Once v8 (YOLOv8, Ultralytics software company, Los Angeles, United States) object detection model with attention mechanism through the Kaggle TBX-11k dataset. The YOLOv8 is used for its ability to detect multiple objects in a single pass. However, it struggles with small objects and finds it impossible to perform fine-grained classifications. To evade this problem, the proposed research incorporates the SFF technique to improve detection performance and decrease small object missed detection rates. Correspondingly, the efficacy of the projected mechanism is calculated utilizing various performance metrics such as recall, precision, F1Score, and mean Average Precision (mAP) to estimate the performance of the proposed framework. Furthermore, the comparison of existing models reveals the efficiency of the proposed research. The present research is envisioned to contribute to the medical world and assist radiologists in identifying tuberculosis using the YOLOv8 model to obtain an optimal outcome.

Introduction: In response to the increasing prevalence of electronic medical records (EMRs) stored in databases, healthcare staff are encountering difficulties retrieving these records due to their limited technical expertise in database operations. As these records are crucial for delivering appropriate medical care, there is a need for an accessible method for healthcare staff to access EMRs.

Methods: To address this, natural language processing (NLP) for Text-to-SQL has emerged as a solution, enabling non-technical users to generate SQL queries using natural language text. This research assesses existing work on Text-to-SQL conversion and proposes the MedT5SQL model specifically designed for EMR retrieval. The proposed model utilizes the Text-to-Text Transfer Transformer (T5) model, a Large Language Model (LLM) commonly used in various text-based NLP tasks. The model is fine-tuned on the MIMICSQL dataset, the first Text-to-SQL dataset for the healthcare domain. Performance evaluation involves benchmarking the MedT5SQL model on two optimizers, varying numbers of training epochs, and using two datasets, MIMICSQL and WikiSQL.

Results: For MIMICSQL dataset, the model demonstrates considerable effectiveness in generating question-SQL pairs achieving accuracy of 80.63%, 98.937%, and 90% for exact match accuracy matrix, approximate string-matching, and manual evaluation, respectively. When testing the performance of the model on WikiSQL dataset, the model demonstrates efficiency in generating SQL queries, with an accuracy of 44.2% on WikiSQL and 94.26% for approximate string-matching.

Discussion: Results indicate improved performance with increased training epochs. This work highlights the potential of fine-tuned T5 model to convert medical-related questions written in natural language to Structured Query Language (SQL) in healthcare domain, providing a foundation for future research in this area.

Semantic segmentation models trained on annotated data fail to generalize well when the input data distribution changes over extended time period, leading to requiring re-training to maintain performance. Classic unsupervised domain adaptation (UDA) attempts to address a similar problem when there is target domain with no annotated data points through transferring knowledge from a source domain with annotated data. We develop an online UDA algorithm for semantic segmentation of images that improves model generalization on unannotated domains in scenarios where source data access is restricted during adaptation. We perform model adaptation by minimizing the distributional distance between the source latent features and the target features in a shared embedding space. Our solution promotes a shared domain-agnostic latent feature space between the two domains, which allows for classifier generalization on the target dataset. To alleviate the need of access to source samples during adaptation, we approximate the source latent feature distribution via an appropriate surrogate distribution, in this case a Gaussian mixture model (GMM).

Persuasive technologies, in connection with human factor engineering requirements for healthy workplaces, have played a significant role in ensuring a change in human behavior. Healthy workplaces suggest different best practices applicable to body posture, proximity to the computer system, movement, lighting conditions, computer system layout, and other significant psychological and cognitive aspects. Most importantly, body posture suggests how users should sit or stand in workplaces in line with best and healthy practices. In this study, we developed two study phases (pilot and main) using two deep learning models: convolutional neural networks (CNN) and Yolo-V3. To train the two models, we collected posture datasets from creative common license YouTube videos and Kaggle. We classified the dataset into comfortable and uncomfortable postures. Results show that our YOLO-V3 model outperformed CNN model with a mean average precision of 92%. Based on this finding, we recommend that YOLO-V3 model be integrated in the design of persuasive technologies for a healthy workplace. Additionally, we provide future implications for integrating proximity detection taking into consideration the ideal number of centimeters users should maintain in a healthy workplace.

Introduction: Hyperdimensional Computing (HDC) is a brain-inspired and lightweight machine learning method. It has received significant attention in the literature as a candidate to be applied in the wearable Internet of Things, near-sensor artificial intelligence applications, and on-device processing. HDC is computationally less complex than traditional deep learning algorithms and typically achieves moderate to good classification performance. A key aspect that determines the performance of HDC is encoding the input data to the hyperdimensional (HD) space.

Methods: This article proposes a novel lightweight approach relying only on native HD arithmetic vector operations to encode binarized images that preserves the similarity of patterns at nearby locations by using point of interest selection and local linear mapping.

Results: The method reaches an accuracy of 97.92% on the test set for the MNIST data set and 84.62% for the Fashion-MNIST data set.

Discussion: These results outperform other studies using native HDC with different encoding approaches and are on par with more complex hybrid HDC models and lightweight binarized neural networks. The proposed encoding approach also demonstrates higher robustness to noise and blur compared to the baseline encoding.

Learning from complex, multidimensional data has become central to computational mathematics, and among the most successful high-dimensional function approximators are deep neural networks (DNNs). Training DNNs is posed as an optimization problem to learn network weights or parameters that well-approximate a mapping from input to target data. Multiway data or tensors arise naturally in myriad ways in deep learning, in particular as input data and as high-dimensional weights and features extracted by the network, with the latter often being a bottleneck in terms of speed and memory. In this work, we leverage tensor representations and processing to efficiently parameterize DNNs when learning from high-dimensional data. We propose tensor neural networks (t-NNs), a natural extension of traditional fully-connected networks, that can be trained efficiently in a reduced, yet more powerful parameter space. Our t-NNs are built upon matrix-mimetic tensor-tensor products, which retain algebraic properties of matrix multiplication while capturing high-dimensional correlations. Mimeticity enables t-NNs to inherit desirable properties of modern DNN architectures. We exemplify this by extending recent work on stable neural networks, which interpret DNNs as discretizations of differential equations, to our multidimensional framework. We provide empirical evidence of the parametric advantages of t-NNs on dimensionality reduction using autoencoders and classification using fully-connected and stable variants on benchmark imaging datasets MNIST and CIFAR-10.

Social media has profoundly changed our modes of self-expression, communication, and participation in public discourse, generating volumes of conversations and content that cover every aspect of our social lives. Social media platforms have thus become increasingly important as data sources to identify social trends and phenomena. In recent years, academics have steadily lost ground on access to social media data as technology companies have set more restrictions on Application Programming Interfaces (APIs) or entirely closed public APIs. This circumstance halts the work of many social scientists who have used such data to study issues of public good. We considered the viability of eight approaches for image-based social media data collection: data philanthropy organizations, data repositories, data donation, third-party data companies, homegrown tools, and various web scraping tools and scripts. This paper discusses the advantages and challenges of these approaches from literature and from the authors' experience. We conclude the paper by discussing mechanisms for improving social media data collection that will enable this future frontier of social science research.