Restoring the patient’s occlusal function of broken teeth is a challenging task since tooth texture is very complex, a slight deviation may affect the patient’s chewing function and temporomandibular joint function. Therefore, how to efficiently repair the complete shape and real surface of the crown is a critical problem. Traditional technologies are hard to restore complete shape of the dental crown or lack inlay surface details, due to dataset limitations and complexity of missing parts. In this paper, we propose a multi-view crown restoration framework MVDC based on contrastive learning. Specifically, MVDC contains: 1) a multi-view generator with a specially designed loss measurement by using contrastive learning; 2) a multi-scale discriminator mechanism able to consider relation and consistency between teeth from different scales; 3) an occlusal groove extraction network to extract the occlusal details. We conducted extensive experiments on existing public datasets. The results showcase the superior performance of MVDC.

The increasing single-cell RNA sequencing (scRNA-seq) data enable researchers to explore cellular heterogeneity and gene expression profiles, offering a high-resolution view of the transcriptome at the single-cell level. However, the dropout events, which are often present in scRNA-seq data, remaining challenges for downstream analysis. Although a number of studies have been developed to recover single-cell expression profiles, their performance may be hindered due to not fully exploring the inherent relations between genes. To address the issue, we propose scDTL, a deep transfer learning based approach for scRNA-seq data imputation by harnessing the bulk RNA-sequencing information. We firstly employ a denoising autoencoder trained on bulk RNA-seq data as the initial imputation model, and then leverage a domain adaptation framework that transfers the knowledge learned by the bulk imputation model to scRNA-seq learning task. In addition, scDTL employs a parallel operation with a 1D U-Net denoising model to provide gene representations of varying granularity, capturing both coarse and fine features of the scRNA-seq data. Finally, we utilize a cross-channel attention mechanism to fuse the features learned from the transferred bulk imputation model and U-Net model. In the evaluation, we conduct extensive experiments to demonstrate that scDTL could outperform other state-of-the-art methods in the quantitative comparison and downstream analyses.

Predicting urban crowd patterns/flows is a challenging task due to complex spatial–temporal (ST) dependencies. In this article, we aim to examine and report the capability and effectiveness of the current most widely used graph convolutional networks (GCNs) on mobile data analysis in ST networks for crowd flow forecasting. Specifically, we propose a novel dual-stream framework leveraging multigraph with attentive ST networks (MG-ASTN) to simultaneously predict crowd in–out flow and origin–destination (OD) flow based on the trajectory data collected by on-board devices (e.g., GPS). MG-ASTN utilizes multi-GCNs encoding non-Euclidean correlations to explore pairwise relationships among regions. In addition, we further apply a cross-channel attention mechanism with 3-D temporal convolutional network to address the heterogeneity of ST features and capture more meaningful data representations for multitask learning. In the evaluation, we conduct experiments based on two real-world data sets and verify most well-known state-of-the-art methods for crowd flow prediction. The results demonstrate that MG-ASTN could outperform other solutions—in–out flow prediction with lowest RMSE and MAE, and OD flow prediction beyond others in most cases, thus it has great potential in modeling the complex correlations among regions in ST networks and enabling accurate prediction in urban computing.

Gesture recognition by using Inertial Measurement Unit (IMU) sensors plays an important role in various Internet of Things (IOT) applications, e.g., smart home, intelligent medical system and so on. Traditional technologies usually utilize machine learning algorithms to train different gestures during the offline phase, then recognize the gesture during the online phase. However, such technologies cannot recognize these gestures without prior training. Even for the same gesture, with different gesture amplitude may result in unsuccessful recognition. Also if we change the person to perform the same gesture, the algorithms fails. In order to overcome these drawbacks, we propose an approach, which will be able to track the human body motion in real-time and also recognize complicated gestures. It utilizes the accelerometer information and proposes comprehensive localization algorithms for each deployed sensor attached on the human body. Then, it takes the correlation and limitation among body parts into account to recognize the gesture. Our experiments results show that, the successful recognition rate of our algorithm is 100%. Furthermore, any part of the human body can be well tracked, the tracking accuracy can reach$0.06m$.

Every year, over 4700 traffic fatalities and 75000 crash injuries involve pedestrians in the United States. Effective solutions are urgently needed to prevent vehicle–pedestrian collision accidents. Many driving assistance systems are proposed to address this problem; however, they require additional infrastructures that may result in higher costs and be difficult to deploy on a large scale. In this article, we propose SafePath, which uses the ubiquitous smartphones to avoid vehicle–pedestrian collision. Specifically, SafePath utilizes the smartphones to broadcast the redesigned service set identifier (SSID) messages containing users’ information (e.g., location, direction, etc.) and scan the surroundings via wireless communications. Considering the limited communication range and the possible interference, and obstruction of obstacles, we propose a collaborative mechanism to enhance the transmission capability, hence predicting the collisions in advance effectively. We also design a risk evaluation scheme to calculate the probability of accidents and inform users to take actions against accidents at different levels. We implement SafePath on the Android platform and conduct extensive real-road experiments to evaluate the system performance. The experimental results demonstrate that SafePath can provide twice the transmission range compared with other collision-avoiding systems. Moreover, it also can significantly reduce the probability of vehicle–pedestrian collisions by up to 81.4%, with respect to other compared collision-avoiding systems in our real-road test.

Flow prediction attracts intensive research interests, since it can offer essential support to many crucial problems in public safety and smart city, e.g., epidemic spread prediction and medical resource allocation optimization. Among all the models in flow prediction, deep learning models (e.g., convolutional neural networks, recurrent neural networks, and graph neural networks) are popular and outperform other statistics and machine learning models, since they can learn intrinsic structures and extract features from spatial–temporal (ST) data. However, most of them set strict temporal periods in the prediction or separate the interaction between spatial and temporal correlations. Therefore, the prediction accuracy is affected. To overcome the difficulties, we propose a flow prediction network attentive spatial–temporal convolutional network (ASTCN), which can effectively handle large-scale flow data and learn complex features. In ASTCN, we leverage an attention mechanism to overcome the previous problem of strict temporal periods, and can effectively fuse ST data with multiple factors from different time-series sources. Furthermore, we propose a causal 3-D convolutional layer based on temporal convolutional networks (TCNs). It can simultaneously extract both spatial and temporal features to improve the prediction accuracy. We comprehensively conducted our experiments based on real-world data sets. Experimental results show that ASTCN outperforms the state-of-the-art methods by at least 3.78% in root mean square error. Therefore, ASTCN is a potential solution to other large-scale ST problems.

Inattentive driving is one of the high-risk factors that causes a large number of traffic accidents every year. In this article, we aim to detect driver inattention leveraging on large-scale vehicle trajectory data while at the same time explore how do these inattentive events affect driver behaviors and what following reactions they may cause, especially, for commercial vehicles. Specifically, the proposed system targets four most commonly occurring critical inattentive events, including smoking, phone call, turning back, and yawning. By applying a deep convolutional neural network (CNN) (Inception v3) with two data augmentation routines—Mixup and synthetic minority oversampling technique (Smote), we are able to balance the training data distribution and improve the generalization of the classification model. Then, based on the output derived from the inattention detection combining with point of interest (POI) and climate data, a long short-term memory (LSTM)-based model is deployed to predict driver upcoming abnormal operations on road (due to inattention) which may result in potential dangerous driving conditions, such as sudden acceleration/deceleration, aggressive left/right lane change, etc. To evaluate our proposed system, we collect more than 120000 real-world driving traces from over 200 drivers. The experimental results show that our model achieves a weight accuracy (WA) of 92.27% for inattentive driving detection and a WA of 91.67% for abnormal driving prediction, demonstrating its great potential of shaping good driving habits and promoting road safety.

Nowadays transceiver-free (also referred to as device-free) localization using Received Signal Strength (RSS) is a hot topic for researchers due to its widespread applicability. However, RSS is easily affected by the indoor environment, resulting in a dense deployment of reference nodes. Some hybrid systems have already been proposed to help RSS localization, but most of them require additional hardware support. In order to solve this problem, in this paper, we propose two algorithms, which leverage the Packet Received Rate (PRR) to help RSS localization without additional hardware support. Moreover, we take the environment noise information into consideration by utilizing the Signal-to-Noise Ratio (SNR) which is based on the RSS and Noise Floor (NF) information instead of pure RSS. Thus, we can alleviate the noise effect in the environment and make our system more sensitive to the target. Specifically, when reference nodes are sparsely deployed and RSS is very weak, PRR and SNR can help in performing localization more accurately. Our BEYOND RSS system is based on sparse wireless sensor networks, wherein the experimental results show that the average localization error of our approach outperforms the pure RSS based approach by about 15.19%.

Emerging blockchain technologies have been increasingly popular and reforming our daily lives. Fusing blockchain technology with existing cloud systems has a great benefits in both improving the functionality/performance and guaranteeing the security/privacy. However, most existing commercial systems fail to address the characteristics of PoS blockchain applications in the cloud. In real-world scenarios, jobs/pods may arrive and leave due to the workload changes. Traditionally, the selection process is based on the state of the workers, e.g., resource availability and specifications of pods. In this paper, we not only provide an optimal solution for offline workloads management which minimizes the number of used workers to reduce the total computational resource demand, but also propose a high performance Kubernetes scheduling scheme HPKS, which maximizes the utilization of workers. Specifically, extensive experiments based on real PoS blockchain applications shows that HPKS reduces the average worker nodes usage by 13.0%. Additionally, the overall increase of Makespan using HPKS is less than 3% when compared to the default scheduler available in Kubernetes.

Driving while distracted or losing alertness significantly increases the risk of traffic accident. The emerging Internet of Things (IoT) systems for smart driving hold the promise of significantly reducing road accidents. In particular, detecting unsafe hand motions and warning the driver using smart sensors can improve the driver’s alertness and skill. However, due to the impact of the vehicle’s movement and the significant variation across different driving environments, detecting the position of the driver’s hand is challenging. This article presents SafeWatch—a system based on smartwatches and smartphones that detects the driver’s unsafe behaviors in a real-time manner. SafeWatch infers driver’s hand position based on several important features, such as the posture of the driver’s forearm and the vibration on the smartwatch. SafeWatch employs a novel adaptive training algorithm that keeps updating the training data set at run-time based on inferred hand positions in certain driving conditions. The evaluation with 75 real driving trips from six subjects shows that SafeWatch has a high accuracy over 97.0% for both recall and precision in detection of the unsafe hand positions when the condition lasts for more than 6.0 s , as well as over 97.1% recall and over 91.0% precision in detection of the unsafe hand movements when it lasts for more than 2.5 s . The relative position of the hand to the steering wheel also reveals some detailed driving habits, like the type of steering method.

Aortic stent has been widely used in restoring vascular stenosis and assisting patients with cardiovascular disease. The effective visualization of aortic stent is considered to be critical to ensure the effectiveness and functions of the aortic stent in clinical practice. Volume rendering with ray casting has been used as an effective approach to enable the effective visualization of aortic stent. The volume rendering relies on the transfer function that converts the medical images into optical attributes including color and transparency. This article proposes a new transfer function, namely, the multi-dimensional transfer function, to provide additional transparency value of a voxel. The proposed approach using the additional transparency value effectively assists the distinguishing of tissues that have the same CT value. The transparency values are simultaneously determined by gray threshold and gray change threshold, which can recognize the unnecessary structures such as bones transparent. A series of experimental results demonstrate that the situation of aorta stent of a patient can be directly observed, and the angle of view can be switched arbitrarily. The proposed method provides a new way for the operation of a virtual endoscopy to reach the place of blood vessels that a traditional endoscopy fails to reach.

Electric vehicles (EVs) play a significant role in the current transportation systems. The main factor that affects the acceptance of existing EV models is the range anxiety problem caused by limited charging stations and long recharge times. Recently, the solar-powered EV has drawn many attentions due to being free of charging limitations. However, the solarpowered EVs may still struggle with the limited use because of unpredictable solar availability. For example, shadings causedby buildings and trees also possibly decrease the solar panel cell efficiency. To address this, we propose a route planning method for solar-powered EVs to balance the energy harvesting and consumption subject to time constraint. The idea behindour solution is to offer power-aware optimal routing, which maximizes the on-road energy input given solar availability on each road segment. We first build a solar access estimation model using 3D geographic data and then employ a multi-criteriasearch method to generate a set of Pareto candidate routes. In order to reduce the size of the set, we leverage the bisect kmeans clustering algorithm to extract the most representative Pareto routes with better solar availability. In the evaluation, wedeveloped a validation platform on the vehicle and leveraged mobile sensing techniques to examine our proposed model in real road environments. We conducted simulations to evaluate our proposed route planning algorithm using real life scenarios. Experimental results demonstrate that our solar input model is robust to real road scenarios, and the routing algorithm has great potential to provide efficient services for solar-powered EV in the future.

A large number of car accidents occur at intersections every year mainly due to drivers' "illegal maneuver" or "unsafe behavior". To promote traffic safety, we present SafeCam, a smartphone-based system that jointly leverages vehicle dynamics and the real-time traffic control information (e.g., traffic signals) to detect and study driver dangerous behaviors at intersections. In particular, SafeCam uses embedded sensors (i.e., inertial sensors) on the phone to generate soft hints tracking different driving conditions while at the same time adopts vision-based algorithms to recognize intersection-related critical driving events including unsafe turns, running stop signs and running red lights. In order to improve the system efficiency, we utilize adaptive color filtering under two lighting conditions (e.g., sunny and cloudy) and deploy the subsampling methods to make a trade off between the detection rate and the processing latency. In the evaluation, we conduct real-road driving experiments involving 15 drivers and 6 vehicles. The experiment results demonstrate that SafeCam is robust and effective in real-road driving environments, and has great potential to alert drivers for their dangerous behaviors at intersections and at the same time help them shape safe driving habits. Our experiments also reveal several interesting findings. 1) On average a driver failed to fully stop at stop signs 3 times in a trip of 3.5 km. 2) 11 out of 15 participants have lane drifting problems when they are making turns in the test. 3) Drivers took longer braking time when they approached a stop sign than a red light.