China
Africa
Explore chapters and articles related to this topic
Wireless Sensor Network Security: A Survey
Published in Yang Xiao, Security in Distributed, Grid, Mobile, and Pervasive Computing, 2007
John Paul Walters, Zhengqiang Liang, Weisong Shi, Vipin Chaudhary
More recently Shrivastava et al. propose a summary structure that is able to support fairly complex aggregate functions, such as median and range queries [75]. It is important to note that typical aggregate functions are capable of performing min/max, sum, and average. The more complex aggregates, such as finding the most frequent data values, are typically not supported. They note that the added aggregate functions are not exact. However, they prove strict guarantees on the approximation quality of the queries [75].
Databases
Published in Ian Foster, Rayid Ghani, Ron S. Jarmin, Frauke Kreuter, Julia Lane, Big Data and Social Science, 2020
The group by operator can be used in conjunction with the aggregate functions to group the result set by one or more columns. For example, we can use the following query to create a table with three columns: investigator name, the number of grants associated with the investigator, and the aggregate funding.
A novel graph neural networks approach for 3D product model retrieval
Published in International Journal of Computer Integrated Manufacturing, 2023
Chengfeng Jian, Yiqiang Lu, Mingliang Lin, Meiyu Zhang
To aggregate these messages, simple sum or average can be used as message aggregate function. However, it is found that some features and geometric information in MBD models do not play a key role in distinguishing models with different nature, and sometimes lead to interference. For example, blend features such as chamfer and tool withdrawal groove are difficult to provide effective identification information, but a large number of blend features are distributed in some models such as ladder shaft, which significantly change the topology structure of the model and the corresponding attribute adjacency graph. Therefore, the existence of blend feature will lead to the phenomenon of missing detection, reduce the recall and precision of the retrieval algorithm, make the retrieval results cannot accurately reflect the user’s retrieval intention, and limit the effective reuse of the existing MBD model design and manufacturing knowledge. In order to solve this problem, the graph attention mechanism [Velickovic et al. (2018)] is introduced to the graph neural network, which can adaptively adjust the weight coefficient of nodes in the computing process according to the structure of attribute adjacency graph. For each pair of nodes, the mechanism takes their hidden state as input and calculates an attention coefficient to indicate the importance of the corresponding neighbor node, which will be formally describe below. Assuming that the current central node is v, and the weight coefficient from w to v of one of the neighbor nodes is:
An aggregate homotopy method for solving unconstrained minimax problems
Published in Optimization, 2021
Based on the linear homotopy and the aggregate function for the objective function, we construct the following aggregate homotopy mapping in this paper: where , is the starting point, is a constant. In the homotopy mapping (9), the homotopy parameter t is also used as the smoothing parameter of the aggregate function. Then, when the homotopy parameter t tends to 0, the smoothing parameter of the aggregate function also tends to 0, and the aggregate function converges to uniformly with respect to . Define the zero point set of as
Data fusion for a GPS/INS tightly coupled positioning system with equality and inequality constraints using an aggregate constraint unscented Kalman filter
Published in Journal of Spatial Science, 2020
Hang Yu, Zengke Li, Jian Wang, Houzeng Han
In this example, we chose the constraint control value and set the controlling factor of the aggregate function . For the UKF and aggre-UKF algorithms, the spreading, contraction and offset parameters were set to 1, 0 and 0, which are the same as used in Simon (2006). The field test trajectories of unconstrained GPS/INS integrated navigation by using the UKF (i.e. scheme 3) are illustrated in Figure 4. As shown in this figure, the trajectory obtained by the UKF deviates from the reference to a certain extent. This can also be concluded from the position errors, calculated by differencing the estimated position states with the reference values, as illustrated in Figure 5, which demonstrates that the errors are at the level of 1–2 m.