EnforceSNN: Enabling Resilient and Energy-Efficient Spiking Neural Network Inference considering Approximate DRAMs for Embedded Systems

Abstract

Spiking Neural Networks (SNNs) have shown capabilities of achieving high accuracy under unsupervised settings and low operational power/energy due to their bio-plausible computations. Previous studies identified that DRAM-based off-chip memory accesses dominate the energy consumption of SNN processing. However, state-of-the-art works do not optimize the DRAM energy-per-access, thereby hindering the SNN-based systems from achieving further energy efficiency gains. To substantially reduce the DRAM energy-per-access, an effective solution is to decrease the DRAM supply voltage, but it may lead to errors in DRAM cells (i.e., so-called approximate DRAM). Towards this, we propose \textit{EnforceSNN}, a novel design framework that provides a solution for resilient and energy-efficient SNN inference using reduced-voltage DRAM for embedded systems. The key mechanisms of our EnforceSNN are: (1) employing quantized weights to reduce the DRAM access energy; (2) devising an efficient DRAM mapping policy to minimize the DRAM energy-per-access; (3) analyzing the SNN error tolerance to understand its accuracy profile considering different bit error rate (BER) values; (4) leveraging the information for developing an efficient fault-aware training (FAT) that considers different BER values and bit error locations in DRAM to improve the SNN error tolerance; and (5) developing an algorithm to select the SNN model that offers good trade-offs among accuracy, memory, and energy consumption. The experimental results show that our EnforceSNN maintains the accuracy (i.e., no accuracy loss for BER less-or-equal 10^-3) as compared to the baseline SNN with accurate DRAM, while achieving up to 84.9\% of DRAM energy saving and up to 4.1x speed-up of DRAM data throughput across different network sizes.

arXiv commentAccepted for publication at Frontiers in Neuroscience - Section Neuromorphic Engineering

References in corpus (17)

• Deep Learning in Spiking Neural Networks
• Deep Learning with Limited Numerical Precision
• Mixed Precision Training
• Quantizing deep convolutional networks for efficient inference: A whitepaper
• A Survey on Edge Computing Systems and Tools
• A 0.086-mm${}^2$ 12.7-pJ/SOP 64k-Synapse 256-Neuron Online-Learning Digital Spiking Neuromorphic Processor in 28nm CMOS
• BindsNET: A machine learning-oriented spiking neural networks library in Python
• MorphIC: A 65-nm 738k-Synapse/mm${}^2$ Quad-Core Binary-Weight Digital Neuromorphic Processor with Stochastic Spike-Driven Online Learning
• SpykeTorch: Efficient Simulation of Convolutional Spiking Neural Networks with at most one Spike per Neuron
• Unsupervised Learning with Self-Organizing Spiking Neural Networks
• FSpiNN: An Optimization Framework for Memory- and Energy-Efficient Spiking Neural Networks
• Understanding Reduced-Voltage Operation in Modern DRAM Chips: Characterization, Analysis, and Mechanisms
• Towards Energy-Efficient and Secure Edge AI: A Cross-Layer Framework
• Stochastic rounding and reduced-precision fixed-point arithmetic for solving neural ordinary differential equations
• ReSpawn: Energy-Efficient Fault-Tolerance for Spiking Neural Networks considering Unreliable Memories
• lpSpikeCon: Enabling Low-Precision Spiking Neural Network Processing for Efficient Unsupervised Continual Learning on Autonomous Agents
• tinySNN: Towards Memory- and Energy-Efficient Spiking Neural Networks

Cited by in corpus (9)

• Embodied Neuromorphic Artificial Intelligence for Robotics: Perspectives, Challenges, and Research Development Stack
• SNN4Agents: A Framework for Developing Energy-Efficient Embodied Spiking Neural Networks for Autonomous Agents
• TopSpark: A Timestep Optimization Methodology for Energy-Efficient Spiking Neural Networks on Autonomous Mobile Agents
• RescueSNN: Enabling Reliable Executions on Spiking Neural Network Accelerators under Permanent Faults
• Continual Learning with Neuromorphic Computing: Foundations, Methods, and Emerging Applications
• SpikeNAS: A Fast Memory-Aware Neural Architecture Search Framework for Spiking Neural Network-based Embedded AI Systems
• Mantis: Enabling Energy-Efficient Autonomous Mobile Agents with Spiking Neural Networks
• FastSpiker: Enabling Fast Training for Spiking Neural Networks on Event-based Data through Learning Rate Enhancements for Autonomous Embedded Systems
• SpikeVox: Towards Energy-Efficient Speech Therapy Framework with Spike-driven Generative Language Models