Search

Predicting computational intensity (CI) is essential for domain decomposition and load balance in parallel geoprocessing. However, traditional CI prediction is limited in capturing the heterogeneity of spatial domain, leading to poor accuracy and load imbalance. Leveraging recent advancements in deep learning from Artificial Intelligence (AI), this paper proposes a deep learning-based approach for predicting CI and enhancing domain decomposition, which reduces the dependency on expert knowledge through automatic feature learning. In the approach, Convolutional Neural Networks are employed to capture the heterogeneity of spatial domain, encompassing structural, distribution, and topological characteristics. A fully connected layer is then utilized for CI prediction and optimized domain decomposition. Comparative experiments were implemented between the proposed approach and three traditional methods, using two cases: spatial intersection on vector data and peak perilousness assessment. The results demonstrate that the proposed approach achieves a speedup ratio of 19.8 and a parallel efficiency of 0.82. The findings highlight the advantages of the proposed approach in terms of parallel performance and usability. This study serves as a valuable reference for illustrating how deep learning can enhance parallel geoprocessing, providing a roadmap for applying deep learning techniques to geocomputations and fostering further advancements of AI GIS.

High performance computing is required for fast geoprocessing of geospatial big data. Using spatial domains to represent computational intensity (CIT) and domain decomposition for parallelism are prominent strategies when designing parallel geoprocessing applications. Traditional domain decomposition is limited in evaluating the computational intensity, which often results in load imbalance and poor parallel performance. From the data science perspective, machine learning from Artificial Intelligence (AI) shows promise for better CIT evaluation. This paper proposes a machine learning approach for predicting computational intensity, followed by an optimized domain decomposition, which divides the spatial domain into balanced subdivisions based on the predicted CIT to achieve better parallel performance. The approach provides a reference framework on how various machine learning methods including feature selection and model training can be used in predicting computational intensity and optimizing parallel geoprocessing against different cases. Some comparative experiments between the approach and traditional methods were performed using the two cases, DEM generation from point clouds and spatial intersection on vector data. The results not only demonstrate the advantage of the approach, but also provide hints on how traditional GIS computation can be improved by the AI machine learning. [ABSTRACT FROM AUTHOR]