Tiny Deep Ensemble: Uncertainty Estimation in Edge AI Accelerators via Ensembling Normalization Layers with Shared Weights

TL;DR

This paper introduces Tiny-Deep Ensemble, a low-cost uncertainty estimation method for edge AI that ensem bles normalization layers with shared weights, significantly reducing hardware overhead while maintaining high accuracy.

Contribution

The paper proposes a novel ensemble approach that ensembles only normalization layers with shared weights, drastically reducing memory and latency overhead for uncertainty estimation on edge devices.

Findings

Reduces latency and memory overhead by up to ~M times.

Maintains or improves inference accuracy by up to ~1%.

Achieves a 17.17% reduction in RMSE across benchmarks.

Abstract

The applications of artificial intelligence (AI) are rapidly evolving, and they are also commonly used in safety-critical domains, such as autonomous driving and medical diagnosis, where functional safety is paramount. In AI-driven systems, uncertainty estimation allows the user to avoid overconfidence predictions and achieve functional safety. Therefore, the robustness and reliability of model predictions can be improved. However, conventional uncertainty estimation methods, such as the deep ensemble method, impose high computation and, accordingly, hardware (latency and energy) overhead because they require the storage and processing of multiple models. Alternatively, Monte Carlo dropout (MC-dropout) methods, although having low memory overhead, necessitate numerous ($\sim 100$) forward passes, leading to high computational overhead and latency. Thus, these approaches are not suitable for battery-powered edge devices with limited computing and memory resources. In this paper, we propose the Tiny-Deep Ensemble approach, a low-cost approach for uncertainty estimation on edge devices. In our approach, only normalization layers are ensembled $M$ times, with all ensemble members sharing common weights and biases, leading to a significant decrease in storage requirements and latency. Moreover, our approach requires only one forward pass in a hardware architecture that allows batch processing for inference and uncertainty estimation. Furthermore, it has approximately the same memory overhead compared to a single model. Therefore, latency and memory overhead are reduced by a factor of up to $\sim M\times$. Nevertheless, our method does not compromise accuracy, with an increase in inference accuracy of up to $\sim 1\%$ and a reduction in RMSE of $17.17\%$ in various benchmark datasets, tasks, and state-of-the-art architectures.