White-Box mHC: Electromagnetic Spectrum-Aware and Interpretable Stream Interactions for Hyperspectral Image Classification

TL;DR

This paper introduces ES-mHC, a transparent hyperspectral image classification model that explicitly models electromagnetic spectrum interactions, enhancing interpretability and providing insights into internal decision mechanisms.

Contribution

We propose a novel hyper-connection framework that explicitly models spectrum interactions, improving interpretability and revealing internal dynamics in hyperspectral image classification.

Findings

Hyper-connection matrices show coherent spatial patterns.

Interaction behaviors are asymmetric and interpretable.

Structured patterns emerge faster with higher expansion rates.

Abstract

In hyperspectral image classification (HSIC), most deep learning models rely on opaque spectral-spatial feature mixing, limiting their interpretability and hindering understanding of internal decision mechanisms. We present physical spectrum-aware white-box mHC, named ES-mHC, a hyper-connection framework that explicitly models interactions among different electromagnetic spectrum groupings (residual stream in mHC) interactions using structured, directional matrices. By separating feature representation from interaction structure, ES-mHC promotes electromagnetic spectrum grouping specialization, reduces redundancy, and exposes internal information flow that can be directly visualized and spatially analyzed. Using hyperspectral image classification as a representative testbed, we demonstrate that the learned hyper-connection matrices exhibit coherent spatial patterns and asymmetric interaction behaviors, providing mechanistic insight into the model internal dynamics. Furthermore, we find that increasing the expansion rate accelerates the emergence of structured interaction patterns. These results suggest that ES-mHC transforms HSIC from a purely black-box prediction task into a structurally transparent, partially white-box learning process.