CloudLSTM: A Recurrent Neural Model for Spatiotemporal Point-cloud Stream Forecasting
This addresses the need for forecasting in geospatial point-cloud data streams, which is incremental as it adapts existing LSTM frameworks to handle non-grid data.
The paper tackles the problem of forecasting spatiotemporal data from point-cloud streams, such as mobile service traffic and air quality indicators, by introducing CloudLSTM with a Dynamic Point-cloud Convolution operator, and it shows that CloudLSTM outperforms competitor models in accurate long-term predictions on real-world datasets.
This paper introduces CloudLSTM, a new branch of recurrent neural models tailored to forecasting over data streams generated by geospatial point-cloud sources. We design a Dynamic Point-cloud Convolution (DConv) operator as the core component of CloudLSTMs, which performs convolution directly over point-clouds and extracts local spatial features from sets of neighboring points that surround different elements of the input. This operator maintains the permutation invariance of sequence-to-sequence learning frameworks, while representing neighboring correlations at each time step -- an important aspect in spatiotemporal predictive learning. The DConv operator resolves the grid-structural data requirements of existing spatiotemporal forecasting models and can be easily plugged into traditional LSTM architectures with sequence-to-sequence learning and attention mechanisms. We apply our proposed architecture to two representative, practical use cases that involve point-cloud streams, i.e., mobile service traffic forecasting and air quality indicator forecasting. Our results, obtained with real-world datasets collected in diverse scenarios for each use case, show that CloudLSTM delivers accurate long-term predictions, outperforming a variety of competitor neural network models.