Figure 2 | Scientific Reports

Figure 2

From: Rapid Learning of Earthquake Felt Area and Intensity Distribution with Real-time Search Engine Queries

Figure 2

A schematic diagram of Query Screening based on Machine Learning. In our approach, every search query \({q}_{i}\) is associated with a weight of confidence \({w}_{i}\), which can be learned by a 3-layer deep neural network (DNN). Specifically, to build the loss function, we assume that if the weights are appropriately estimated, the weighted average (i.e., centroid) of the query locations can be used to estimate the epicenters of earthquakes.

Back to article page