Towards Energy-Efficient and Secure Edge AI: A Cross-Layer Framework

Abstract

The security and privacy concerns along with the amount of data that is required to be processed on regular basis has pushed processing to the edge of the computing systems. Deploying advanced Neural Networks (NN), such as deep neural networks (DNNs) and spiking neural networks (SNNs), that offer state-of-the-art results on resource-constrained edge devices is challenging due to the stringent memory and power/energy constraints. Moreover, these systems are required to maintain correct functionality under diverse security and reliability threats. This paper first discusses existing approaches to address energy efficiency, reliability, and security issues at different system layers, i.e., hardware (HW) and software (SW). Afterward, we discuss how to further improve the performance (latency) and the energy efficiency of Edge AI systems through HW/SW-level optimizations, such as pruning, quantization, and approximation. To address reliability threats (like permanent and transient faults), we highlight cost-effective mitigation techniques, like fault-aware training and mapping. Moreover, we briefly discuss effective detection and protection techniques to address security threats (like model and data corruption). Towards the end, we discuss how these techniques can be combined in an integrated cross-layer framework for realizing robust and energy-efficient Edge AI systems.

arXiv commentTo appear at the 40th IEEE/ACM International Conference on Computer-Aided Design (ICCAD), November 2021, Virtual Event

References in corpus (31)

• Neural Architecture Search with Reinforcement Learning
• Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding
• Deep Learning with Limited Numerical Precision
• A 0.086-mm${}^2$ 12.7-pJ/SOP 64k-Synapse 256-Neuron Online-Learning Digital Spiking Neuromorphic Processor in 28nm CMOS
• CryptoDL: Deep Neural Networks over Encrypted Data
• Hardware and Software Optimizations for Accelerating Deep Neural Networks: Survey of Current Trends, Challenges, and the Road Ahead
• Towards the Science of Security and Privacy in Machine Learning
• I Know What You See: Power Side-Channel Attack on Convolutional Neural Network Accelerators
• Chiron: Privacy-preserving Machine Learning as a Service
• ALWANN: Automatic Layer-Wise Approximation of Deep Neural Network Accelerators without Retraining
• Robust Machine Learning Systems: Challenges, Current Trends, Perspectives, and the Road Ahead
• Terminal Brain Damage: Exposing the Graceless Degradation in Deep Neural Networks Under Hardware Fault Attacks
• FSpiNN: An Optimization Framework for Memory- and Energy-Efficient Spiking Neural Networks
• Q-SpiNN: A Framework for Quantizing Spiking Neural Networks
• DeepAPT: Nation-State APT Attribution Using End-to-End Deep Neural Networks
• NASCaps: A Framework for Neural Architecture Search to Optimize the Accuracy and Hardware Efficiency of Convolutional Capsule Networks
• ReSpawn: Energy-Efficient Fault-Tolerance for Spiking Neural Networks considering Unreliable Memories
• DRMap: A Generic DRAM Data Mapping Policy for Energy-Efficient Processing of Convolutional Neural Networks
• Securing Deep Spiking Neural Networks against Adversarial Attacks through Inherent Structural Parameters
• Is Spiking Secure? A Comparative Study on the Security Vulnerabilities of Spiking and Deep Neural Networks
• ROMANet: Fine-Grained Reuse-Driven Off-Chip Memory Access Management and Data Organization for Deep Neural Network Accelerators
• QuSecNets: Quantization-based Defense Mechanism for Securing Deep Neural Network against Adversarial Attacks
• SparkXD: A Framework for Resilient and Energy-Efficient Spiking Neural Network Inference using Approximate DRAM
• NeuroAttack: Undermining Spiking Neural Networks Security through Externally Triggered Bit-Flips
• TrISec: Training Data-Unaware Imperceptible Security Attacks on Deep Neural Networks
• Q-CapsNets: A Specialized Framework for Quantizing Capsule Networks
• CapsAttacks: Robust and Imperceptible Adversarial Attacks on Capsule Networks
• DESCNet: Developing Efficient Scratchpad Memories for Capsule Network Hardware
• ReD-CaNe: A Systematic Methodology for Resilience Analysis and Design of Capsule Networks under Approximations
• PANDA: Facilitating Usable AI Development
• A Low-cost Fault Corrector for Deep Neural Networks through Range Restriction

Cited by in corpus (10)

• EnforceSNN: Enabling Resilient and Energy-Efficient Spiking Neural Network Inference considering Approximate DRAMs for Embedded Systems
• Embodied Neuromorphic Artificial Intelligence for Robotics: Perspectives, Challenges, and Research Development Stack
• lpSpikeCon: Enabling Low-Precision Spiking Neural Network Processing for Efficient Unsupervised Continual Learning on Autonomous Agents
• A Homomorphic Encryption Framework for Privacy-Preserving Spiking Neural Networks
• RoHNAS: A Neural Architecture Search Framework with Conjoint Optimization for Adversarial Robustness and Hardware Efficiency of Convolutional and Capsule Networks
• 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
• A Methodology to Study the Impact of Spiking Neural Network Parameters considering Event-Based Automotive Data
• Special Session: Towards an Agile Design Methodology for Efficient, Reliable, and Secure ML Systems
• SpikeVox: Towards Energy-Efficient Speech Therapy Framework with Spike-driven Generative Language Models