Layer-Wise Modality Decomposition for Interpretable Multimodal Sensor Fusion

2025/11/02 by Jaehyun Park, Park, Jaehyun, Konyul Park +7
Computer Science · #Advanced Neural Network Applications #Adversarial Robustness in Machine Learning #Computer Vision and Pattern Recognition (cs.CV) #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences

paper · pdf · doi:10.48550/arxiv.2511.00859

openalex publication_date 2025/11/02 · openalex created_date 2025/11/06 · openalex updated_date 2026/07/28

Abstract

In autonomous driving, transparency in the decision-making of perception models is critical, as even a single misperception can be catastrophic. Yet with multi-sensor inputs, it is difficult to determine how each modality contributes to a prediction because sensor information becomes entangled within the fusion network. We introduce Layer-Wise Modality Decomposition (LMD), a post-hoc, model-agnostic interpretability method that disentangles modality-specific information across all layers of a pretrained fusion model. To our knowledge, LMD is the first approach to attribute the predictions of a perception model to individual input modalities in a sensor-fusion system for autonomous driving. We evaluate LMD on pretrained fusion models under camera-radar, camera-LiDAR, and camera-radar-LiDAR settings for autonomous driving. Its effectiveness is validated using structured perturbation-based metrics and modality-wise visual decompositions, demonstrating practical applicability to interpreting high-capacity multimodal architectures. Code is available at https://github.com/detxter-jvb/Layer-Wise-Modality-Decomposition.

Citations

• Layer-Wise Relevance Propagation with Conservation Property for ResNet
• RCBEVDet: Radar-camera Fusion in Bird's Eye View for 3D Object Detection
• Gaussian Process Neural Additive Models
• AttnLRP: Attention-Aware Layer-Wise Relevance Propagation for Transformers
• CRN: Camera Radar Net for Accurate, Robust, Efficient 3D Perception
• Simple-BEV: What Really Matters for Multi-Sensor BEV Perception?
• Neural Basis Models for Interpretability
• BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representation
• XAI for Transformers: Better Explanations through Conservative Propagation
• NODE-GAM: Neural Generalized Additive Model for Interpretable Deep Learning
• Transformer Interpretability Beyond Attention Visualization
• Analyzing the Source and Target Contributions to Predictions in Neural\n Machine Translation
• Deformable DETR: Deformable Transformers for End-to-End Object Detection
• Neural Additive Models: Interpretable Machine Learning with Neural Nets
• Purifying Interaction Effects with the Functional ANOVA: An Efficient Algorithm for Recovering Identifiable Additive Models
• Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networks
• nuScenes: A multimodal dataset for autonomous driving
• Local Rule-Based Explanations of Black Box Decision Systems
• A Unified Approach to Interpreting Model Predictions
• Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization
• Layer Normalization
• Training Deep Nets with Sublinear Memory Cost
• Layer-wise Relevance Propagation for Neural Networks with Local\n Renormalization Layers
• Deep Residual Learning for Image Recognition
• U-Net: Convolutional Networks for Biomedical Image Segmentation
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

Related

• FusionAD: Multi-modality Fusion for Prediction and Planning Tasks of Autonomous Driving
• Multi-modal Sensor Fusion for Auto Driving Perception: A Survey
• Integrating Multi-Modal Sensors: A Review of Fusion Techniques for Intelligent Vehicles
• t-READi: Transformer-Powered Robust and Efficient Multimodal Inference for Autonomous Driving
• Towards Sensor Data Abstraction of Autonomous Vehicle Perception Systems