Device-aware inference operations in SONOS nonvolatile memory arrays

04/02/2020

∙

Non-volatile memory arrays can deploy pre-trained neural network models for edge inference. However, these systems are affected by device-level noise and retention issues. Here, we examine damage caused by these effects, introduce a mitigation strategy, and demonstrate its use in fabricated array of SONOS (Silicon-Oxide-Nitride-Oxide-Silicon) devices. On MNIST, fashion-MNIST, and CIFAR-10 tasks, our approach increases resilience to synaptic noise and drift. We also show strong performance can be realized with ADCs of 5-8 bits precision.

READ FULL TEXT

Device-aware inference operations in SONOS nonvolatile memory arrays

Non-Volatile Memory Array Based Quantization- and Noise-Resilient LSTM Neural Networks

A Sequential Concept Drift Detection Method for On-Device Learning on Low-End Edge Devices

Synaptic metaplasticity with multi-level memristive devices

Optimizing Neural Network for Computer Vision task in Edge Device

Deep Neural Network Optimized to Resistive Memory with Nonlinear Current-Voltage Characteristics

CIMulator: A Comprehensive Simulation Platform for Computing-In-Memory Circuit Macros with Low Bit-Width and Real Memory Materials

Network insensitivity to parameter noise via adversarial regularization

Device-aware inference operations in SONOS nonvolatile memory arrays

Related Research

Non-Volatile Memory Array Based Quantization- and Noise-Resilient LSTM Neural Networks

A Sequential Concept Drift Detection Method for On-Device Learning on Low-End Edge Devices

Synaptic metaplasticity with multi-level memristive devices

Optimizing Neural Network for Computer Vision task in Edge Device

Deep Neural Network Optimized to Resistive Memory with Nonlinear Current-Voltage Characteristics

CIMulator: A Comprehensive Simulation Platform for Computing-In-Memory Circuit Macros with Low Bit-Width and Real Memory Materials

Network insensitivity to parameter noise via adversarial regularization