Inferring new facts from an existing knowledge graph with explainable reasoning processes is an important problem, known as knowledge graph (KG) reasoning. The problem is often formulated as finding ...the specific path that represents the query relation and connects the query entity and the correct answer. However, due to the limited expressiveness of individual paths, the majority of previous works failed to capture the complex subgraph structure in the graph. We propose CogKR that traverses the knowledge graph to conduct multi-hop reasoning. More specifically, motivated by the dual process theory from cognitive science, our framework is composed of an extension module and a reasoning module. By setting up a cognitive graph through iteratively coordinating the two modules, CogKR can cope with more complex reasoning scenarios in the form of subgraphs instead of individual paths. Experiments on three knowledge graph reasoning benchmarks demonstrate that CogKR achieves significant improvements in accuracy compared with previous methods while providing the explainable capacity. Moreover, we evaluate CogKR on the challenging one-shot link prediction task, exhibiting the superiority of the framework on accuracy and scalability compared to the state-of-the-art approaches.
Detection of dynamic triggering of earthquakes combined with microseismic monitoring allows a better understanding on rock damage process in rock engineering, which involve local stress disturbances ...caused by far-field earthquakes. However, rapid and efficient detection of dynamic triggering of earthquakes remain clueless due to the lack of enough investigation. This study proposed a novel method for automatic detection of dynamic triggering of earthquakes based on convolutional neural networks (CNNs). Results show that the trained model is capable of detecting earthquakes with a high accuracy rate of 98.3%, which provided strong supports for the automatic detection of dynamic triggering of earthquakes. This method is applied to the Xiaojiang fault region, and dynamically triggered earthquakes were spotted for 31 of the 47 selected earthquakes. Our method achieves 94% recognition accuracy on the detection of dynamic triggering of earthquakes. This method offers a rapid and accurate framework to detect dynamic triggering of earthquakes and investigate the damage to rock engineering structures caused by far-field earthquakes based on microseismic monitoring.
•This study proposed a novel method for automatic detection of dynamic triggering based on convolutional neural networks.•The proposed method obtained from the training can be applied to different regional seismic networks.•For detection of dynamic triggering, our method can achieve a 94% correct identification rate.
Federated Distillation (FD) extends classic Federated Learning (FL) to a more general training framework that enables model-heterogeneous collaborative learning by Knowledge Distillation (KD) across ...multiple clients and the server. However, existing KD-based algorithms usually require a set of shared input samples for each client to produce soft-prediction for distillation. Worse still, such a manual selection is accompanied by careful deliberations or prior information on clients' private data distribution, which is not in line with the privacy-preserving characteristic of classic FL. In this paper, we propose a novel training framework to achieve data-independent knowledge transfer by properly designing a distributed generative adversarial network (GAN) between the server and clients that can synthesize shared feature representations to facilitate the FD training. Specifically, we deploy a generator on the server and reuse each local model as a federated discriminator to form a lightweight efficient distributed GAN that can automatically synthesize simulated global feature representations for distillation. Moreover, since the synthesized feature representations are usually more faithful and homologous with global data distribution, faster and better training convergence can be obtained. Extensive experiments on different tasks and heterogeneous models demonstrate the effectiveness of the proposed framework on model accuracy and communication overhead.
Graph neural networks (GNNs) are popularly used to analyze non-euclidean graph data. Despite their successes, the design of graph neural networks requires heavy manual work and rich domain knowledge. ...Recently, neural architecture search algorithms are widely used to automatically design neural architectures for CNNs and RNNs. Inspired by the success of neural architecture search algorithms, we present a graph neural architecture search algorithm GraphNAS that enables automatic design of the best graph neural architecture based on reinforcement learning. Specifically, GraphNAS uses a recurrent network as the controller to generate variable-length strings that describe the architectures of graph neural networks, and trains the recurrent network with policy gradient to maximize the expected accuracy of the generated architectures on a validation data set. Moreover, based on GraphNAS, we design a new GraphNAS++ model using distributed neural architecture search. Compared with GraphNAS that generates and evaluates only one candidate architecture at each iteration, GraphNAS++ generates a mini-batch of candidate architectures and evaluates them in a distributed computing environment until convergence. Experiments on real-world graph datasets demonstrate that GraphNAS can design a novel network architecture that rivals the best human-invented architecture in terms of accuracy. Moreover, GraphNAS++ can speed up the design process at least five times by using the distributed training framework with GPUs.
Influenced by the great success of deep learning via cloud computing and the rapid development of edge chips, research in artificial intelligence (AI) has shifted to both of the computing paradigms, ...i.e., cloud computing and edge computing. In recent years, we have witnessed significant progress in developing more advanced AI models on cloud servers that surpass traditional deep learning models owing to model innovations (e.g., Transformers, Pretrained families), explosion of training data and soaring computing capabilities. However, edge computing, especially edge and cloud collaborative computing, are still in its infancy to announce their success due to the resource-constrained IoT scenarios with very limited algorithms deployed. In this survey, we conduct a systematic review for both cloud and edge AI. Specifically, we are the first to set up the collaborative learning mechanism for cloud and edge modeling with a thorough review of the architectures that enable such mechanism. We also discuss potentials and practical experiences of some on-going advanced edge AI topics including pretraining models, graph neural networks and reinforcement learning. Finally, we discuss the promising directions and challenges in this field.
•There is a strong correlation between concrete damage degree and erosion mass loss.•A DV threshold for the step surge of concrete cavitation erosion is suggested.•An incentive surge mechanism of ...concrete cavitation erosion due to damage exists.
Flood releasing safety is one of the three major safety problems of high dams due to the high-water head and large discharge, and the safety problems caused by cavitation erosion are increasingly prominent, which has become one of the challenging problems for the safety of high dam flood discharge. Particularly, flood discharge and energy dissipation structures are susceptible to varying degrees of damage under external loading and material performance degradation. However, the action mechanism of damage-cavitation erosion for concrete remains unknown and unclear. In this paper, the cavitation erosion properties of concrete with different damage variables and the influence mechanism of damage-cavitation erosion are systematically studied. Based on the ultrasonic cavitation erosion tests of undamaged and damaged concrete, it is found that the mass loss, harshness, volume of pits, and depth of damaged concrete increased rapidly at first and then slowly, the maximum loss rate is almost dependent on the degree of damage, and its relationship with the damage degree can be described by a piecewise linear function. More importantly, a damage variable threshold for the step surge of concrete cavitation erosion is suggested. The cementitious material in damaged concrete is more prone to cavitation erosion due to the potential microcracks caused by the damage. Compared with the results of tests and simulations, the incentive mechanism of the surge of concrete cavitation erosion due to the concrete damage - microcrack initiation - strength and fatigue life reduction - microjet impact - water wedge excitation in cracks - development, expansion, and penetration of erosion pits is revealed. The influence mechanism revelation of damage-cavitation erosion for concrete will enrich the theory and technology of cavitation erosion risk assessment for hydraulic concrete structures.
Loess landslides are widely distributed in Northern China and pose a significant threat to human life, natural resources, and infrastructure in mountainous regions. Accurate segmentation and ...measurement of loess landslides is crucial to documenting their occurrence and extent and investigating the distribution, types, and patterns of slope failures. The measurement of landslides also assists in assessing their susceptibility and risk. Herein, a novel loess landslide segmentation and measurement framework based on deep learning is proposed. Multiple experts label the ground-truth landslide regions, and simultaneous truth and performance level estimation (STAPLE) masks are generated. The U-Net segmentation algorithm is trained using a supervised approach to segment the loess landslide region. The fully connected conditional random field is integrated into the U-Net to further optimize the segmentation quality. In the final step, the predicted landslide boundaries are visualized, and the diameters (e.g., length and width) of the segmentation outcome are computed simultaneously. Four state-of-the-art segmentation algorithms are selected for the comparative analysis. The computational results demonstrate that the proposed framework outperforms all the other algorithms tested in terms of segmentation accuracy and boundary errors. The results verify the advantages of using STAPLE and U-Net integrated with a conditional random field.