CA-Stream: Attention-based pooling for interpretable image recognition

2024/04/23 by Felipe Torres, Torres, Felipe, Hanwei Zhang +7 · 1 citation
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.2404.14996

openalex publication_date 2024/04/23 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Explanations obtained from transformer-based architectures in the form of raw attention, can be seen as a class-agnostic saliency map. Additionally, attention-based pooling serves as a form of masking the in feature space. Motivated by this observation, we design an attention-based pooling mechanism intended to replace Global Average Pooling (GAP) at inference. This mechanism, called Cross-Attention Stream (CA-Stream), comprises a stream of cross attention blocks interacting with features at different network depths. CA-Stream enhances interpretability in models, while preserving recognition performance.

Cited by

• Robust Noise Attenuation via Adaptive Pooling of Transformer Outputs

Related

• Stochastic Region Pooling: Make Attention More Expressive
• Mix-Pooling Strategy for Attention Mechanism
• CLUENet: Cluster Attention Makes Neural Networks Have Eyes
• Augmenting Convolutional networks with attention-based aggregation
• Artifacts and Attention Sinks: Structured Approximations for Efficient Vision Transformers