HiRo: A Compact Four-Directional Hierarchical Reservoir Token-Mixer for Efficient Image Classification (opens in new tab)

Recent image classification models must balance local feature modeling, cross-window interaction, and parameter efficiency. Many high-performing architectures rely on fully trainable token-mixers, which improve representation learning but increase parameter count, optimization complexity and computational cost. We propose a parameter-efficient image classification model called HiRo that integrates shifted-window partitioning with multi-direction...