An Analysis of Temporal Dropout in Earth Observation Time Series for Regression Tasks
This addresses uncertainty quantification for satellite data analysis in Earth Observation, but it is incremental as it builds on existing dropout and Monte Carlo methods.
The paper tackles the problem of missing time-steps in Earth Observation time series data for regression tasks, which degrades predictive performance, by introducing Monte Carlo Concrete Temporal Dropout (MC-ConcTD) to learn optimal dropout distributions during inference, improving predictive performance and uncertainty calibration on three datasets.
Missing instances in time series data impose a significant challenge to deep learning models, particularly in regression tasks. In the Earth Observation field, satellite failure or cloud occlusion frequently results in missing time-steps, introducing uncertainties in the predicted output and causing a decline in predictive performance. While many studies address missing time-steps through data augmentation to improve model robustness, the uncertainty arising at the input level is commonly overlooked. To address this gap, we introduce Monte Carlo Temporal Dropout (MC-TD), a method that explicitly accounts for input-level uncertainty by randomly dropping time-steps during inference using a predefined dropout ratio, thereby simulating the effect of missing data. To bypass the need for costly searches for the optimal dropout ratio, we extend this approach with Monte Carlo Concrete Temporal Dropout (MC-ConcTD), a method that learns the optimal dropout distribution directly. Both MC-TD and MC-ConcTD are applied during inference, leveraging Monte Carlo sampling for uncertainty quantification. Experiments on three EO time-series datasets demonstrate that MC-ConcTD improves predictive performance and uncertainty calibration compared to existing approaches. Additionally, we highlight the advantages of adaptive dropout tuning over manual selection, making uncertainty quantification more robust and accessible for EO applications.