English

Feature-level Interaction Explanations in Multimodal Transformers

Machine Learning 2026-03-17 v1 Artificial Intelligence

Abstract

Multimodal Transformers often produce predictions without clarifying how different modalities jointly support a decision. Most existing multimodal explainable AI (MXAI) methods extend unimodal saliency to multimodal backbones, highlighting important tokens or patches within each modality, but they rarely pinpoint which cross-modal feature pairs provide complementary evidence (synergy) or serve as reliable backups (redundancy). We present Feature-level I2MoE (FL-I2MoE), a structured Mixture-of-Experts layer that operates directly on token/patch sequences from frozen pretrained encoders and explicitly separates unique, synergistic, and redundant evidence at the feature level. We further develop an expert-wise explanation pipeline that combines attribution with top-K% masking to assess faithfulness, and we introduce Monte Carlo interaction probes to quantify pairwise behavior: the Shapley Interaction Index (SII) to score synergistic pairs and a redundancy-gap score to capture substitutable (redundant) pairs. Across three benchmarks (MMIMDb, ENRICO, and MMHS150K), FL-I2MoE yields more interactionspecific and concentrated importance patterns than a dense Transformer with the same encoders. Finally, pair-level masking shows that removing pairs ranked by SII or redundancy-gap degrades performance more than masking randomly chosen pairs under the same budget, supporting that the identified interactions are causally relevant.

Keywords

Cite

@article{arxiv.2603.13326,
  title  = {Feature-level Interaction Explanations in Multimodal Transformers},
  author = {Yeji Kim and Housam Khalifa Bashier Babiker and Mi-Young Kim and Randy Goebel},
  journal= {arXiv preprint arXiv:2603.13326},
  year   = {2026}
}
R2 v1 2026-07-01T11:19:02.094Z