Exploring Self-Attention for Crop-type Classification Explainability

TL;DR

This paper evaluates the explainability of transformer attention weights in crop-type classification, revealing their correlation with phenological events but also their limitations in characterizing crop phenology.

Contribution

It introduces a novel framework for systematically assessing the explanatory power of attention weights in crop classification models.

Findings

Attention patterns relate to key phenological dates.

Attention weights have limited capability to characterize crop phenology.

Sensitivity analysis shows dependence on training crop set.

Abstract

Transformer models have become a promising approach for crop-type classification. Although their attention weights can be used to understand the relevant time points for crop disambiguation, the validity of these insights depends on how closely the attention weights approximate the actual workings of these black-box models, which is not always clear. In this paper, we introduce a novel explainability framework that systematically evaluates the explanatory power of the attention weights of a standard transformer encoder for crop-type classification. Our results show that attention patterns strongly relate to key dates, which are often associated with critical phenological events for crop-type classification. Further, the sensitivity analysis reveals the limited capability of the attention weights to characterize crop phenology as the identified phenological events depend on the other crops considered during training. This limitation highlights the relevance of future work towards the development of deep learning approaches capable of automatically learning the temporal vegetation dynamics for accurate crop disambiguation