Process-Guided Concept Bottleneck Model

TL;DR

The paper introduces PG-CBM, an extension of Concept Bottleneck Models that incorporates domain-specific causal mechanisms, improving accuracy, interpretability, and scientific insight in Earth Observation biomass estimation.

Contribution

PG-CBM integrates domain-defined causal constraints into CBMs, enabling better performance and interpretability with sparse supervision in scientific applications.

Findings

Reduces error and bias in biomass estimation

Leverages multi-source heterogeneous data effectively

Enhances transparency and scientific insights

Abstract

Concept Bottleneck Models (CBMs) improve the explainability of black-box Deep Learning (DL) by introducing intermediate semantic concepts. However, standard CBMs often overlook domain-specific relationships and causal mechanisms, and their dependence on complete concept labels limits applicability in scientific domains where supervision is sparse but processes are well defined. To address this, we propose the Process-Guided Concept Bottleneck Model (PG-CBM), an extension of CBMs which constrains learning to follow domain-defined causal mechanisms through biophysically meaningful intermediate concepts. Using above ground biomass density estimation from Earth Observation data as a case study, we show that PG-CBM reduces error and bias compared to multiple benchmarks, whilst leveraging multi-source heterogeneous training data and producing interpretable intermediate outputs. Beyond improved accuracy, PG-CBM enhances transparency, enables detection of spurious learning, and provides scientific insights, representing a step toward more trustworthy AI systems in scientific applications.