Learning the low-dimensional representations of graphs (i.e., network embedding) plays a critical role in network analysis and facilitates many downstream tasks. Recently graph convolutional networks (GCNs) have revolutionized the field of network embedding, and led to state-of-the-art performance in network analysis tasks such as link prediction and node classification. Nevertheless, most of the existing GCN-based network embedding methods are proposed for unsigned networks. However, in the real world, some of the networks are signed, where the links are annotated with different polarities, e.g., positive vs. negative. Since negative links may have different properties from the positive ones and can also significantly affect the quality of network embedding. Thus in this paper, we propose a novel network embedding framework SNEA to learn Signed Network Embedding via graph Attention. In particular, we propose a masked self-attentional layer, which leverages self-attention mechanism to estimate the importance coefficient for pair of nodes connected by different type of links during the embedding aggregation process. Then SNEA utilizes the masked self-attentional layers to aggregate more important information from neighboring nodes to generate the node embeddings based on balance theory. Experimental results demonstrate the effectiveness of the proposed framework through signed link prediction task on several real-world signed network datasets.
Adaptive closed-loop control is used to optimize the lift-to-drag ratio of post-stall separated flow over a NACA 0025 airfoil using multiple amplitudemodulated (AM) or burst-modulated (BM), zero-net-mass-flux (ZNMF) actuators that cover the central 33% span of the airfoil. A simplex optimization approach uses the lift and drag measured by a strain-gauge balance for feedback and searches for the optimal AM or BM actuation parameters in a closed-loop fashion. An energy penalty function based on the electrical power consumption of the actuators is added to the cost function to study the tradeoff between the aerodynamic performance and the power requirement. In the baseline post-stall separated flow over the airfoil, two dominant characteristic frequencies are identified via flow visualization and hot-wire anemometry: the convective instability of the separated shear layer and the global instability of the vortex shedding in the wake. Closedloop control increases the lift-to-drag ratio by a factor of 2-3 via smallamplitude AM and BM forcing of nonlinear interactions between these instabilities. I.
In this paper, we introduce a new reinforcement learning (RL) based neural architecture search (NAS) methodology for effective and efficient generative adversarial network (GAN) architecture search. The key idea is to formulate the GAN architecture search problem as a Markov decision process (MDP) for smoother architecture sampling, which enables a more effective RL-based search algorithm by targeting the potential global optimal architecture. To improve efficiency, we exploit an off-policy GAN architecture search algorithm that makes efficient use of the samples generated by previous policies. Evaluation on two standard benchmark datasets (i.e., CIFAR-10 and STL-10) demonstrates that the proposed method is able to discover highly competitive architectures for generally better image generation results with a considerably reduced computational burden: 7 GPU hours. Our code is available at https://github.com/Yuantian013/E2GAN.
The older and deeper hydrocarbon accumulations receive increasing attention across the world, providing more technical and commercial challenges to hydrocarbon exploration. We present a study of an asymmetrical, N-S striking intracratonic sag which developed across the Sichuan basin, south China, from Late Ediacaran to Early Cambrian times. The Mianyang-Changning intracratonic sag is ∼50 km wide, with its steepest part in the basin center. In particular the eastern margin shows its greatest steepness. Five episodes in the evolutions of the sag can be recognized. It begins in the Late Ediacaran with an uplift and erosion correlated to Tongwan movement. Initial extension occurred during the Early Cambrian Maidiping period, when more strata of the Maidiping Formation were deposited across the sag. Subsequently, maximum extension occurred during the Early Cambrian Qiongzhusi period that resulted in 450-1700 m thick Maidiping-Canglangpu Formations being deposited in the sag. Then, the sag disappeared at the Longwangmiao period, as it was infilled by the sediments. The intracratonic sag has significant influence on the development of high-quality reservoirs in the Dengying and Longwangmiao Formations and source-rock of the Niutitang Formation. It thus indicates that a high probability for oil/gas accumulation exists along the intracratonic sag, across the central Sichuan basin.
Movement is a complex process that evolves through both space and time. Movement data generated by moving objects is a kind of big data, which has been a focus of research in science, technology, economics, and social studies. Movement database is also at the forefront of geographic information science research. Developing efficient access methods for movement data stored in movement databases is of critical importance. Tree-like indexing structures such as the R-tree, Quadtree, Octree are not suitable for indexing multi-dimensional movement data because they all have high space cost of their inner nodes. In addition, it is difficult to use them for parallel access to multidimensional movement data because they thereof, are in hierarchical structures, which have severe overlapping problems in high dimensional space. In this paper, we propose a novel access method, the Decomposition Tree (D-tree), for indexing multi-dimensional movement data. The D-tree is a virtual tree without inner nodes, instead, through an encoding method based on integer bit-shifting operation, and can efficiently answer a wide range of queries. Experimental results show that the space cost and query performance of D-tree are superior to its best known competitors.
As the autonomous vehicles technology gradually enters the public eye, understanding consumers' psychological motivations for accepting autonomous vehicles is critical for the development of autonomous vehicles and society. Previously, researchers have explored the determinants of fully autonomous vehicles but the relevant research is far from enough. Moreover, the relationship between anthropomorphism and users' behavior has been ignored to a large extent. Therefore, this study aim to fill the gap by using anthropomorphism and the unified theory of acceptance and use of technology (UTAUT) to explore how system attributes (i.e., perceived anthropomorphism, perceived intelligence) and UTAUT attributes influence consumers' acceptance behavior. The data were collected via questionnaire survey conducted in Beijing, China, which can be a promising early adopter of AVs. Structural equation modeling was used to analyze the data. The results reveal that perceived anthropomorphism and perceived intelligence have a direct positive influence on the adoption of AVs; performance expectancy, effort expectancy, and facilitating conditions have an indirect positive influence on intention to adopt AVs. Also, this research contributes to the literature by enriching studies on psychological determinants of autonomous vehicles' adoption by taking an initial step to highlight anthropomorphism perceptions. This can provide managerial implications for policy-makers and businesses on how to effectively allocate resources to enhance autonomous vehicle adoption.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.