Litcius/Paper detail

Evaluations on Deep Neural Networks Training Using Posit Number System

Jinming Lu, Chao Fang, Mingyang Xu, Jun Lin, Zhongfeng Wang

2020IEEE Transactions on Computers68 citationsDOI

Abstract

The training of Deep Neural Networks (DNNs) brings enormous memory requirements and computational complexity, which makes it a challenge to train DNN models on resource-constrained devices. Training DNNs with reduced-precision data representation is crucial to mitigate this problem. In this article, we conduct a thorough investigation on training DNNs with low-bit posit numbers, a Type-III universal number (Unum). Through a comprehensive analysis of quantization with various data formats, it is demonstrated that the posit format shows great potential to be employed in the training of DNNs. Moreover, a DNN training framework using 8-bit posit is proposed with a novel tensor-wise scaling scheme. The experiments show the same performance as the state-of-the-art (SOTA) across multiple datasets (MNIST, CIFAR-10, ImageNet, and Penn Treebank) and model architectures (LeNet-5, AlexNet, ResNet, MobileNet-V2, and LSTM). We further design an energy-efficient hardware prototype for our framework. Compared to the standard floating-point counterpart, our design achieves a reduction of 68, 51, and 75 percent in terms of area, power, and memory capacity, respectively.

Topics & Concepts

MNIST databaseTreebankComputer scienceDeep neural networksQuantization (signal processing)Artificial neural networkArtificial intelligenceScheme (mathematics)Computer engineeringMachine learningAlgorithmDependency (UML)MathematicsMathematical analysisAdvanced Neural Network ApplicationsTensor decomposition and applicationsComputational Physics and Python Applications