Bit Efficient Quantization for Deep Neural Networks
This work addresses efficient inference for edge devices, but it appears incremental as it compares existing quantization methods.
The paper tackles the problem of reducing memory and power usage for deep neural networks on edge devices by comparing model-parameter driven quantization approaches, achieving as low as 3-bit precision without accuracy loss on datasets like ImageNet.
Quantization for deep neural networks have afforded models for edge devices that use less on-board memory and enable efficient low-power inference. In this paper, we present a comparison of model-parameter driven quantization approaches that can achieve as low as 3-bit precision without affecting accuracy. The post-training quantization approaches are data-free, and the resulting weight values are closely tied to the dataset distribution on which the model has converged to optimality. We show quantization results for a number of state-of-art deep neural networks (DNN) using large dataset like ImageNet. To better analyze quantization results, we describe the overall range and local sparsity of values afforded through various quantization schemes. We show the methods to lower bit-precision beyond quantization limits with object class clustering.