🤖 AI Summary
This work addresses the instability and high computational and memory overhead commonly caused by conflicting aggregation strategies in ensemble methods. For the first time, it identifies this conflict as a critical issue and introduces a training-free, test-time adaptive aggregation mechanism that dynamically optimizes the fusion of multiple model outputs without requiring additional training or parameter updates. The proposed approach substantially enhances prediction stability and efficiency, achieving over 96% Top-1 accuracy on ImageNet-1K. Notably, it enables lightweight models to outperform Vision Transformers (ViTs) by 11.75% in accuracy while reducing model parameters by more than 95% and cutting computational costs by 89.1%.
📝 Abstract
Ensemble methods are widely used to improve predictive performance, but their effectiveness often comes at the cost of increased memory usage and computational complexity. In this paper, we identify a conflict in aggregation strategies that negatively impacts prediction stability. We propose StableTTA, a training-free method to improve aggregation stability and efficiency. Empirical results on ImageNet-1K show gains of 10.93--32.82\% in top-1 accuracy, with 33 models achieving over 95\% accuracy and several surpassing 96\%. Notably, StableTTA allows lightweight architectures to outperform ViT by 11.75\% in top-1 accuracy while using less than 5\% of parameters and reducing computational cost by approximately 89.1\% (in GFLOPs), enabling high-accuracy inference on resource-constrained devices.