Pneumonia mainly refers to lung infections caused by pathogens, such as bacteria and viruses. Currently, deep learning methods have been applied to identify pneumonia. However, the traditional deep learning methods for pneumonia identification take less account of the influence of the lung X-ray image background on the model's testing effect, which limits the improvement of the model's accuracy. In this paper, we propose a deep learning method that considers image background factors and analyzes the proposed method with explainable deep learning for explainability. The essential idea is to remove the image background, improve the pneumonia recognition accuracy, and apply the Grad-CAM method to obtain an explainable deep learning model for pneumonia identification. In the proposed approach, (1) preliminary deep learning models for pneumonia X-ray image identification without considering the background are built; (2) deep learning models for pneumonia X-ray image identification with background consideration are built to improve the accuracy of pneumonia identification; (3) Grad-CAM method is employed to analyze the explainability. The proposed approach improves the accuracy of pneumonia identification, and the highest accuracy of VGG16 reaches 95.6%. The proposed approach can be applied to real pneumonia identification for early detection and treatment.

The availability of Internet of Things (IoT) data is essential for the operation of intelligent systems, such as smart energy systems. Unfortunately, information sensitivity and the lack of observations tend to impact the availability of IoT data. To solve this problem, this article proposes an attention-based cycle-consistent generative adversarial network (ABC-GAN) to generate IoT data. By efficiently learning the distribution among different data patterns and sufficiently capturing temporal features, ABC-GAN can effectively reproduce the IoT data collected from different devices and regions. Various experimental results in smart energy systems demonstrate that ABC-GAN excels at capturing the temporal features, distribution, and latent manifolds of the original data when compared to the baselines and that the prediction models trained with the data generated by ABC-GAN can achieve performances similar to models trained with the real data.

Deep learning technology has become one of the core driving forces to promote the in-depth development of industrial automation. In [A1], Wang et al. interpreted the decision process of the convolutional neural network (CNN) by constructing a percolation model from a statistical physics perspective. In this perspective, the decision-making basis of CNN is difficult to understand, because CNN is usually used as a black box model. Furthermore, a novel concept of the differentiation degree and summarized an empirical formula for quantifying the differentiation degree is presented and discussed.

As a vital research subject in the field of intelligent transportation systems (ITSs), traffic flow prediction using deep learning methods has attracted much attention in recent years. However, numerous existing studies mainly focus on short-term traffic flow predictions and fail to consider the influence of external factors. Effective long-term traffic flow prediction has become a challenging issue. As a solution to these challenges, this paper proposes a deep learning approach based on a spatiotemporal graph convolutional network for long-term traffic flow prediction with multiple factors. In the proposed method, our innovative idea is to introduce an attribute feature unit (AF-unit) to fuse external factors into a spatiotemporal graph convolutional network. The proposed method consists of (1) constructing a weighted adjacency matrix using Gaussian similarity functions; (2) assembling a feature matrix to store time-series traffic flow; (3) building an external attribute matrix composed of external factors, including temperature, visibility, and weather conditions; and (4) building a spatiotemporal graph convolutional network based on a deep learning architecture (i.e., T-GCN). The experimental results indicate that (1) the performance of our method considering spatiotemporal dependence has better prediction capability than baseline models; (2) the fusion of meteorological factors can reduce the inaccuracy of traffic prediction; and (3) our method has high accuracy and stability in long-term traffic flow prediction.

Summary Taxi GPS trajectories can be mined and used to optimize urban traffic scheduling. The optimization of traffic scheduling is important, especially in megacities such as Beijing. In this paper, we analyze the traffic features in Beijing by mining taxi GPS trajectories. We define the Congestion Coefficient of each edge of the taxi trajectory as the consumed time of a taxi running over a unit of distance. By analyzing the distribution of congestion coefficients of all taxi trajectories, we can observe that, on working days, (1) the congestion coefficient is between 0 and 2 (average speed is greater than 0.5 m/s) and is acceptable to taxi drivers, (2) the morning rush hours are 7:00 ∼ 10:00, (3) the evening rush hours are 17:00 ∼ 20:00, and (4) the traffic congestion in the morning rush hours is worse than that in the evening rush hours; on the weekend, (1) the congestion coefficient is less than 0.2 (average speed is greater than 5 m/s) and is acceptable to taxi drivers; (2) compared with the traffic congestion on working days, there are no significant morning rush hours on the weekend; and (3) the period of the time between 13:00 and 15:00 could be considered the traffic rush hours on the weekend. These findings can be used to improve urban traffic management.

Summary The mining of human mobility can be exploited to support the design of traffic planning, route recommendations, urban planning, emergency management, and land use. Currently, various methods such as the machine learning algorithms, statistical methods, and semantic analysis are widely applied to identify and extract human mobility patterns. In this paper, we propose a simple and generic paradigm for mining human mobility patterns based on the GPS trajectory data using complex network analysis. The essential ideas behind the proposed paradigm mainly include (1) creating weighted complex networks of GPS trajectories and (2) extracting the human mobility patterns by analyzing the structures and metrics of the created complex networks of GPS trajectories. To evaluate the performance of the proposed paradigm, we design five groups of experiments and identify the mobility patterns of the selected five persons based on the selected five network analysis metrics. Experimental results indicate that (1) the proposed paradigm is effective and (2) the quite interesting potential information about the human mobility can be mined easily. The proposed paradigm is simple and generic, which can be employed to rapidly identify the human mobility patterns based on the GPS trajectory data.

Abstract Landslides are one of the most critical categories of natural disasters worldwide and induce severely destructive outcomes to human life and the overall economic system. To reduce its negative effects, landslides prevention has become an urgent task, which includes investigating landslide-related information and predicting potential landslides. Machine learning is a state-of-the-art analytics tool that has been widely used in landslides prevention. This paper presents a comprehensive survey of relevant research on machine learning applied in landslides prevention, mainly focusing on (1) landslides detection based on images, (2) landslides susceptibility assessment, and (3) the development of landslide warning systems. Moreover, this paper discusses the current challenges and potential opportunities in the application of machine learning algorithms for landslides prevention.

Abstract Slope deformation prediction is crucial for early warning of slope failure, which can prevent property damage and save human life. Existing predictive models focus on predicting the displacement of a single monitoring point based on time series data, without considering spatial correlations among monitoring points, which makes it difficult to reveal the displacement changes in the entire monitoring system and ignores the potential threats from nonselected points. To address the above problem, this paper presents a novel deep learning method for predicting the slope deformation, by considering the spatial correlations between all points in the entire displacement monitoring system. The essential idea behind the proposed method is to predict the slope deformation based on the global information (i.e., the correlated displacements of all points in the entire monitoring system), rather than based on the local information (i.e., the displacements of a specified single point in the monitoring system). In the proposed method, (1) a weighted adjacency matrix is built to interpret the spatial correlations between all points, (2) a feature matrix is assembled to store the time-series displacements of all points, and (3) one of the state-of-the-art deep learning models, i.e., T-GCN, is developed to process the above graph-structured data consisting of two matrices. The effectiveness of the proposed method is verified by performing predictions based on a real dataset. The proposed method can be applied to predict time-dependency information in other similar geohazard scenarios, based on time-series data collected from multiple monitoring points.

Summary An effective strategy for accelerating the calculation of convex hulls is to filter the input points by discarding interior points. In this paper, we present such a straightforward preprocessing approach by discarding the points locating in a convex polygon formed by 16 extreme points. Extreme points of a planar point set do not alter when all points are rotated with the same angle in the plane. Four groups of four extreme points with min or max x or y coordinates can be found for the original point set and three rotated point sets. These 16 extreme points are used to form a planar convex polygon. We discard those points locating in the convex polygon and calculate the desired convex hull of the remaining points. The proposed preprocessing algorithm is evaluated on two computational platforms. Experiments show that, when employing the proposed preprocessing algorithm on the computational platform 1, it achieves speedups of approximately 4 ×∼5× on average and 5 ×∼6× in the best cases over the cases where the proposed approach is not used, while on the computational platform 2, the speedups are approximately 6 ×∼9× on average and 9 ×∼14× in the best cases. Moreover, more than 99% input points can be discarded in most cases.

Geologic hazards (geohazards) are naturally occurring or human-activity-induced geologic conditions capable of causing damage or loss of property and/or life. geohazards, such as landslides, surface subsidence, and earthquakes, can seriously affect and threaten life, property, or public safety. geohazards prevention is the application of geologic engineering principles and existing and emerging technologies to reduce, minimize, or prevent the effects of various geologic hazards. Monitoring and early warning are the most common strategies for geohazards prevention. With the development of the Internet of Things (IoT), an emerging new idea is to apply IoT technology to enhance the accuracy and efficiency of monitoring and early warning systems for geohazards prevention. This article aims to present a comprehensive survey of relevant research and technological developments of the IoT applied in geohazards prevention. It first surveys the applications of the IoT in the monitoring and early warning of seven types of common geohazards, including landslides, debris flow, rockfall, surface subsidence, surface collapse, surface cracks, and earthquakes, then investigates the key technologies in geohazards prevention when utilizing the IoT, and finally summarizes the challenges in IoT-based monitoring and early warning systems for geohazards prevention. Moreover, this article also highlights the future directions for employing the IoT for geohazards prevention.

An effective strategy for accelerating the calculation of convex hulls for planar point sets is to filter the input points by discarding interior points. In this paper, we present such a straightforward preprocessing approach by exploiting the modern Graphics Processing Unit. The basic idea behind our approach is to discard the points that locate inside a convex polygon formed by 16 extreme points. Due to the fact that the extreme points of a planar point set do not alter when all points are rotated with the same angle in the plane, four groups of extreme points with min or max x or y coordinates can be found in the original point set and three rotated point sets. These 16 extreme points are then used to form a planar convex polygon. We check all input points and discard the points that locate inside the convex polygon, and then use the remaining points to calculate the desired convex hull. Experimental results show that: when employing the proposed preprocessing algorithm, it achieves the speedups of about 4x ~ 5x on average and 5x ~ 6x in the best cases over the cases where the proposed approach is not used. In addition, more than 99% input points can be discarded in most experimental tests.

Summary This paper focuses on designing and implementing parallel Moving Least Squares (MLS) interpolation algorithms by exploiting the Graphics Processing Unit (GPU) for the usage in Meshfree methods. The MLS method is an approach for scattered points' approximation / interpolation, which is commonly employed as the shape functions in various Meshfree methods. To improve the computational efficiency in building stiffness matrices in Meshfree methods, we are specifically interested in parallelizing the MLS interpolation on the GPU. The accelerated Meshfree methods can be employed to numerically analyze large deformations of soil and rock masses such as the landslides. In this paper, we first introduce our previously proposed method for finding the k nearest neighboring points located within the local region of the interest point in MLS. We then develop one sequential and three parallel implementations for each of three variations of the MLS interpolation, including the serial implementation, the parallel implementation on the multi‐core CPU, the parallel implementation on a single GPU, and the parallel implementation on multi‐GPUs. To evaluate the computational performance of our GPU implementations, five groups of benchmark tests are conducted in both two‐dimensions and three‐dimensions. We observe that our GPU implementations can achieve satisfied speedups over the sequential CPU implementation for varied sizes of testing data. To benefit the community, all source code and testing data related to the presented parallel MLS interpolation are publicly available.

Post-traumatic stress disorder (PTSD) is a traumatic-stressor related disorder developed by exposure to a traumatic or adverse environmental event that caused serious harm or injury. Structured interview is the only widely accepted clinical practice for PTSD diagnosis but suffers from several limitations including the stigma associated with the disease. Diagnosis of PTSD patients by analyzing speech signals has been investigated as an alternative since recent years, where speech signals are processed to extract frequency features and these features are then fed into a classification model for PTSD diagnosis. In this paper, we developed a deep belief network (DBN) model combined with a transfer learning (TL) strategy for PTSD diagnosis. We computed three categories of speech features and utilized the DBN model to fuse these features. The TL strategy was utilized to transfer knowledge learned from a large speech recognition database, TIMIT, for PTSD detection where PTSD patient data is difficult to collect. We evaluated the proposed methods on two PTSD speech databases, each of which consists of audio recordings from 26 patients. We compared the proposed methods with other popular methods and showed that the state-of-the-art support vector machine (SVM) classifier only achieved an accuracy of 57.68%, and TL strategy boosted the performance of the DBN from 61.53% to 74.99%. Altogether, our method provides a pragmatic and promising tool for PTSD diagnosis.

To minimize the number of birdstrikes, a common method is to use microphone arrays to monitor and identify dangerous birds near the airport or some critical locations in the airspace. However, it was recognized that the range of existing ground-based acoustic monitoring devices is only limited to a few hundred meters. Moreover, the bird classification performance in low signal-to-noise environments such as airports is not very satisfactory. This paper summarizes the development of a high performance bird classification system using a hidden Markov model (HMM) and Gaussian mixture model (GMM). Experimental results verified the classification performance.

Nonlinear amplifiers, present in both the base stations of cellular networks and transponders of satellite communications systems, are known to be one of the most undesired hardware distortions in the future DS/CDMA PCS systems. To mitigate the effects of nonlinearities, this paper proposes a design which modifies the input value of the correlator by employing a discriminator. We derive the optimal function of the discriminator. By computer simulation, we know we can improve the performance using the discriminator.