A Latent Space Metric for Enhancing Prediction Confidence in Earth Observation Data
Ioannis Pitsiorlas, Argyro Tsantalidou, George Arvanitakis, Marios Kountouris, Charalambos Kontoes
TL;DR
The paper tackles predicting mosquito abundance from Earth Observation data and enhancing trust in those predictions by estimating a confidence metric. It introduces a latent-space-based approach using a Variational AutoEncoder, where the encoder maps inputs to latent variables and the decoder produces MA estimates; the confidence metric is defined so that it is proportional to the expected prediction error in latent space. A key finding is a measurable correlation of approximately $0.46$ between the absolute prediction error and the proposed confidence metric, indicating that latent-space distance reflects prediction reliability. The methodology offers a transferable framework for reliability assessment in EO-driven health analytics, with potential to improve decision-making under data-limited conditions across two European AOIs (Veneto, Italy and Upper Rhine Valley, Germany).
Abstract
This study presents a new approach for estimating confidence in machine learning model predictions, specifically in regression tasks utilizing Earth Observation (EO) data, with a particular focus on mosquito abundance (MA) estimation. We take advantage of a Variational AutoEncoder architecture, to derive a confidence metric by the latent space representations of EO datasets. This methodology is pivotal in establishing a correlation between the Euclidean distance in latent representations and the Absolute Error (AE) in individual MA predictions. Our research focuses on EO datasets from the Veneto region in Italy and the Upper Rhine Valley in Germany, targeting areas significantly affected by mosquito populations. A key finding is a notable correlation of 0.46 between the AE of MA predictions and the proposed confidence metric. This correlation signifies a robust, new metric for quantifying the reliability and enhancing the trustworthiness of the AI model's predictions in the context of both EO data analysis and mosquito abundance studies.