Approximate Multiplier

Approximate multipliers are low-power, computationally efficient alternatives to precise multipliers, primarily investigated for their potential to improve the energy efficiency and robustness of deep neural networks (DNNs). Current research focuses on integrating these multipliers into DNN architectures through methods like differentiable neural architecture search (DARTS) and exploring their impact on DNN accuracy under various adversarial attacks. This research is significant because it addresses the high computational cost of DNNs, particularly relevant for resource-constrained applications like edge computing, while also potentially enhancing their resilience to malicious inputs.