Exploring Distance-Aware Uncertainty Quantification for Remote Sensing Image Classification

Deep Learning models for classification often suffer from overconfidence, which naturally results in poor predictive uncertainty estimates. To overcome this, many calibration techniques have been established. These techniques operate on the labels or the output space of the network but ignore the in...

Full description

Saved in:
Bibliographic Details
Main Authors: Koller, Christoph, Jung, Peter, Zhu, Xiao Xiang
Format: Conference Proceeding
Language:English
Subjects:
Calibration
Classification
Deep learning
Image classification
Land Use
OoD Detection
Out-of-Distribution (OoD)
Predictive models
Remote sensing
Residual Network
Spectral Normalization
Uncertainty
Uncertainty Quantification
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Deep Learning models for classification often suffer from overconfidence, which naturally results in poor predictive uncertainty estimates. To overcome this, many calibration techniques have been established. These techniques operate on the labels or the output space of the network but ignore the input image space. A recently proposed approach considers the distances between different network inputs explicitly and theoretically propagates the distances through the network. The resulting predictive uncertainties of the model are then able to better reflect these distances. We test this approach in the context of remote sensing image classification for land use. To evaluate the predictive uncertainties, we set up an Out-of Distribution (OoD) detection framework based on class separation.
ISSN:2153-7003
DOI:10.1109/IGARSS52108.2023.10281435