PikeLPN: Mitigating Overlooked Inefficiencies of Low-Precision Neural Networks

Marina Neseem, Conor McCullough, Randy Hsin, Chas Leichner, Shan Li, In Suk Chong, Andrew Howard, Lukasz Lew, Sherief Reda, Ville-Mikko Rautio, Daniele Moro; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 15996-16005

Low-precision quantization is recognized for its efficacy in neural network optimization. Our analysis reveals that non-quantized elementwise operations which are prevalent in layers such as parameterized activation functions batch normalization and quantization scaling dominate the inference cost of low-precision models. These non-quantized elementwise operations are commonly overlooked in SOTA efficiency metrics such as Arithmetic Computation Effort (ACE). In this paper we propose ACEv2 - an extended version of ACE which offers a better alignment with the inference cost of quantized models and their energy consumption on ML hardware. Moreover we introduce PikeLPN a model that addresses these efficiency issues by applying quantization to both elementwise operations and multiply-accumulate operations. In particular we present a novel quantization technique for batch normalization layers named QuantNorm which allows for quantizing the batch normalization parameters without compromising the model performance. Additionally we propose applying Double Quantization where the quantization scaling parameters are quantized. Furthermore we recognize and resolve the issue of distribution mismatch in Separable Convolution layers by introducing Distribution-Heterogeneous Quantization which enables quantizing them to low-precision. PikeLPN achieves Pareto-optimality in efficiency-accuracy trade-off with up to 3X efficiency improvement compared to SOTA low-precision models.