Artificial intelligence (AI) coupled with promising machine learning (ML) techniques well known from computer science is broadly affecting many aspects of various fields including science and technology, industry, and even our day-to-day life. The ML techniques have been developed to analyze high-throughput data with a view to obtaining useful insights, categorizing, predicting, and making evidence-based decisions in novel ways, which will promote the growth of novel applications and fuel the sustainable booming of AI. This paper undertakes a comprehensive survey on the development and application of AI in different aspects of fundamental sciences, including information science, mathematics, medical science, materials science, geoscience, life science, physics, and chemistry. The challenges that each discipline of science meets, and the potentials of AI techniques to handle these challenges, are discussed in detail. Moreover, we shed light on new research trends entailing the integration of AI into each scientific discipline. The aim of this paper is to provide a broad research guideline on fundamental sciences with potential infusion of AI, to help motivate researchers to deeply understand the state-of-the-art applications of AI-based fundamental sciences, and thereby to help promote the continuous development of these fundamental sciences.

The vehicle impact velocity and vehicle front-end shape are 2 dominant factors that influence the pedestrian kinematics and injury severity. A significant reduction of all injuries can be achieved for all vehicle types when the vehicle impact velocity is less than 30 km/h. Vehicle designs consisting of a short front-end and a wide windshield area can protect pedestrians from fatalities. The results also could be valuable in the design of a pedestrian-friendly vehicle front-end shape. [Supplementary materials are available for this article. Go to the publisher's online edition of Traffic Injury Prevention for the following free supplemental resource: Head impact conditions and injury parameters in four-type vehicle collisions and validation result of the finite element model of one-box vehicle and minicar. ].

WebVR, a multi-user online virtual reality engine, is introduced. The main contributions are mapping the geographical space and virtual space to the P2P overlay network space, and dividing the three spaces by quad-tree method. The geocoding is identified with Hash value, which is used to index the user list, terrain data, and the model object data. Sharing of data through improved Kademlia network model is designed and implemented. In this model, XOR algorithm is used to calculate the distance of the virtual space. The model greatly improves the hit rate of 3D geographic data search under P2P overlay network. Some data preprocessing methods have been adopted to accelerate the data transfer. 3D Global data is used for testing the engine. The test result indicates that, without considering the client bandwidth limit, the more users, the faster loading.

In this study, we investigate the effects of modelling choices for the brain-skull interface (layers of tissues between the brain and skull that determine boundary conditions for the brain) and the constitutive model of brain parenchyma on the brain responses under violent impact as predicted using computational biomechanics model. We used the head/brain model from Total HUman Model for Safety (THUMS)-extensively validated finite element model of the human body that has been applied in numerous injury biomechanics studies. The computations were conducted using a well-established nonlinear explicit dynamics finite element code LS-DYNA. We employed four approaches for modelling the brain-skull interface and four constitutive models for the brain tissue in the numerical simulations of the experiments on post-mortem human subjects exposed to violent impacts reported in the literature. The brain-skull interface models included direct representation of the brain meninges and cerebrospinal fluid, outer brain surface rigidly attached to the skull, frictionless sliding contact between the brain and skull, and a layer of spring-type cohesive elements between the brain and skull. We considered Ogden hyperviscoelastic, Mooney-Rivlin hyperviscoelastic, neo-Hookean hyperviscoelastic and linear viscoelastic constitutive models of the brain tissue. Our study indicates that the predicted deformations within the brain and related brain injury criteria are strongly affected by both the approach of modelling the brain-skull interface and the constitutive model of the brain parenchyma tissues. The results suggest that accurate prediction of deformations within the brain and risk of brain injury due to violent impact using computational biomechanics models may require representation of the meninges and subarachnoidal space with cerebrospinal fluid in the model and application of hyperviscoelastic (preferably Ogden-type) constitutive model for the brain tissue.

Predicting the passenger flow of metro networks is of great importance for traffic management and public safety. However, such predictions are very challenging, as passenger flow is affected by complex spatial dependencies (nearby and distant) and temporal dependencies (recent and periodic). In this paper, we propose a novel deep-learning-based approach, named STGCNNmetro (spatiotemporal graph convolutional neural networks for metro), to collectively predict two types of passenger flow volumes-inflow and outflow-in each metro station of a city. Specifically, instead of representing metro stations by grids and employing conventional convolutional neural networks (CNNs) to capture spatiotemporal dependencies, STGCNNmetro transforms the city metro network to a graph and makes predictions using graph convolutional neural networks (GCNNs). First, we apply stereogram graph convolution operations to seamlessly capture the irregular spatiotemporal dependencies along the metro network. Second, a deep structure composed of GCNNs is constructed to capture the distant spatiotemporal dependencies at the citywide level. Finally, we integrate three temporal patterns (recent, daily, and weekly) and fuse the spatiotemporal dependencies captured from these patterns to form the final prediction values. The STGCNNmetro model is an end-to-end framework which can accept raw passenger flow-volume data, automatically capture the effective features of the citywide metro network, and output predictions. We test this model by predicting the short-term passenger flow volume in the citywide metro network of Shanghai, China. Experiments show that the STGCNNmetro model outperforms seven well-known baseline models (LSVR, PCA-kNN, NMF-kNN, Bayesian, MLR, M-CNN, and LSTM). We additionally explore the sensitivity of the model to its parameters and discuss the distribution of prediction errors.

In this work, we propose an introspection technique for deep neural networks that relies on a generative model to instigate salient editing of the input image for model interpretation. Such modification provides the fundamental interventional operation that allows us to obtain answers to counterfactual inquiries, i.e., what meaningful change can be made to the input image in order to alter the prediction. We demonstrate how to reveal interesting properties of the given classifiers by utilizing the proposed introspection approach on both the MNIST and the CelebA dataset.

Recently, the spatio-temporal residual network (ST-ResNet) has leveraged the power of deep learning (DL) to predict citywide spatio-temporal flow volume.However, this model, neglects the dynamic dependency of the input series in the temporal dimension, which affects the captured spatio-temporal features. The present study introduces the long short-term memory (LSTM) neural network into ST-ResNet, to form a hybrid integrated DL model for citywide spatio-temporal flow volume prediction (called HIDLST). The new model is capable of dynamically learning the temporal dependency via the feedback connection of the LSTM, improving the accuracy of the spatio-temporal features. We test the HIDLST model by predicting the citywide taxi flow volume in Beijing, China. We tune the hyperparameters of the HIDLST model to optimize the prediction accuracy. Comparative experiments indicate that the proposed model consistently outperforms ST-ResNet and several other typical DL-based models with regard to prediction accuracy. Additionally, we also discuss the distribution of prediction errors and the contributions of different spatio-temporal patterns.

No abstract