INSTA-BNN: Binary Neural Network with INSTAnce-aware Threshold

TL;DR

INSTA-BNN introduces an instance-aware threshold mechanism for binary neural networks, dynamically adjusting quantization thresholds based on input statistics to improve accuracy with minimal overhead.

Contribution

The paper proposes a novel dynamic threshold adjustment method for BNNs using higher-order statistics, enhancing accuracy without significant computational cost.

Findings

Outperforms baseline by 3.0% and 2.8% on ImageNet

Achieves 68.5% and 72.2% top-1 accuracy on ResNet-18 and MobileNetV1

Maintains comparable computational cost

Abstract

Binary Neural Networks (BNNs) have emerged as a promising solution for reducing the memory footprint and compute costs of deep neural networks, but they suffer from quality degradation due to the lack of freedom as activations and weights are constrained to the binary values. To compensate for the accuracy drop, we propose a novel BNN design called Binary Neural Network with INSTAnce-aware threshold (INSTA-BNN), which controls the quantization threshold dynamically in an input-dependent or instance-aware manner. According to our observation, higher-order statistics can be a representative metric to estimate the characteristics of the input distribution. INSTA-BNN is designed to adjust the threshold dynamically considering various information, including higher-order statistics, but it is also optimized judiciously to realize minimal overhead on a real device. Our extensive study shows that INSTA-BNN outperforms the baseline by 3.0% and 2.8% on the ImageNet classification task with comparable computing cost, achieving 68.5% and 72.2% top-1 accuracy on ResNet-18 and MobileNetV1 based models, respectively.