Bitwidth-Adaptive Quantization-Aware Neural Network Training: A Meta-Learning Approach

Abstract

Deep neural network quantization with adaptive bitwidths has gained increasing attention due to the ease of model deployment on various platforms with different resource budgets. In this paper, we propose a meta-learning approach to achieve this goal. Specifically, we propose MEBQAT, a simple yet effective way of bitwidth-adaptive quantization-aware training (QAT) where meta-learning is effectively combined with QAT by redefining meta-learning tasks to incorporate bitwidths. After being deployed on a platform, MEBQAT allows the (meta-)trained model to be quantized to any candidate bitwidth with minimal inference accuracy drop. Moreover, in a few-shot learning scenario, MEBQAT can also adapt a model to any bitwidth as well as any unseen target classes by adding conventional optimization or metric-based meta-learning. We design variants of MEBQAT to support both (1) a bitwidth-adaptive quantization scenario and (2) a new few-shot learning scenario where both quantization bitwidths and target classes are jointly adapted. Our experiments show that merging bitwidths into meta-learning tasks results in remarkable performance improvement: 98.7% less storage cost compared to bitwidth-dedicated QAT and 94.7% less back propagation compared to bitwidth-adaptive QAT in bitwidth-only adaptation scenarios, while improving classification accuracy by up to 63.6% compared to vanilla meta-learning in bitwidth-class joint adaptation scenarios.