Bayesian Deep Learning for Hyperspectral Image Classification With Low Uncertainty

In recent years, deep learning models have been widely used for hyperspectral image (HSI) classification, and most of the existing deep-learning-based methods merely focused on high classification accuracy. However, in real applications, classification with low uncertainty matters as much as accurate classification. Unfortunately, the existing methods fail to consider uncertainty. To tackle this challenge, for the first time, Bayesian deep learning (BDL) is investigated to analyze the model uncertainty for HSI classification. Specifically, first, at the feature extraction (FE) stage, an HSI classification framework based on BDL, which contains two Bayesian Gabor layers and a global pooling layer (i.e., BDL-G222), is proposed. In BDL-G222, parameters in Gabor layers are sampled from the Gaussian distribution. The proposed BDL-G222 not only provides the uncertainty estimation but also strengthens the structure characteristic (i.e., texture) of HSI. Second, to model the uncertainty at the final classification stage, BDL-G222 is combined with a Bayesian fully connected layer (BFL) (i.e., BDL-G222-BFL), where the parameters' distribution is adjusted adaptively. In the proposed BDL-G222-BFL, the uncertainty at FE and classification stages is captured, and a whole uncertainty estimation framework is established. Experimental results on the three public HSI datasets demonstrate the superiority in both accuracy and uncertainty. The proposed BDL-based methods pioneer a new direction and provide useful inspiration and experience for practical applications.

Automated summary

In recent years, deep learning models have been widely used for hyperspectral image (HSI) classification. However, most existing methods only focus on high classification accuracy and ignore uncertainty. To address this issue, Bayesian deep learning (BDL) is investigated to analyze the model uncertainty for HSI classification. Experimental results on public HSI datasets demonstrate the superiority of the proposed BDL-based methods in both accuracy and uncertainty estimation.