Variation Aware Training of Hybrid Precision Neural Networks with 28nm HKMG FeFET Based Synaptic Core

TL;DR

This paper introduces a hybrid-precision neural network training method utilizing FeFET-based synaptic devices and SRAM for efficient computation and weight updates, achieving high accuracy despite device variations.

Contribution

It presents a novel hybrid training framework combining FeFET memory and SRAM, enabling robust neural network training with 28nm devices under variation effects.

Findings

Achieved up to 95% inference accuracy with device variations.

Demonstrated robustness of the hybrid training approach.

Validated performance using calibrated device models.

Abstract

This work proposes a hybrid-precision neural network training framework with an eNVM based computational memory unit executing the weighted sum operation and another SRAM unit, which stores the error in weight update during back propagation and the required number of pulses to update the weights in the hardware. The hybrid training algorithm for MLP based neural network with 28 nm ferroelectric FET (FeFET) as synaptic devices achieves inference accuracy up to 95% in presence of device and cycle variations. The architecture is primarily evaluated using behavioral or macro-model of FeFET devices with experimentally calibrated device variations and we have achieved accuracies compared to floating-point implementations.