Bits-Ensemble: Towards Light-Weight Robust Deep Ensemble by Bits-Sharing

Robustness and uncertainty estimation are crucial to the safety of deep neural networks deployed on the edge. The deep ensemble model, composed of a set of individual deep neural networks (namely members), has strong performance in accuracy, uncertainty estimation, and robustness to out-of-distribution data and adversarial attacks. However, the storage and memory consumption increases linearly with the number of members within an ensemble. Previous works focus on selecting better members, layer-wise low-rank approximation of ensemble parameters and designing partial ensemble model for reducing the ensemble size, thus lowering storage and memory consumption. In this work, we pay attention to the quantization of the ensemble, which serves as the last mile of network deployment. We propose a differentiable and parallelizable bit sharing scheme that allows the members to share the less significant bits of parameters, without hurting the performance, leaving alone the more significant bits. The intuition is that, numerically, more significant bits (e.g., the bit for the sign) are more useful in distinguishing a member from other members. For real deployment of the bit-sharing scheme, we further propose an efficient encoding-decoding scheme with minimal storage overhead. Experimental results show that, BitsEnsemble reduces the storage size of ensemble for over 22×, with only 0.36× increase in training latency, and no sacrifice of inference latency. The code is available in https://github.com/ralphc1212/bitsensemble.