Background and Related Work
Alberto Marchisio, Muhammad Shafique · 2024
This book proposes cross-layer optimizations for advanced DL architectures, covering energy efficiency and resiliency perspectives. Besides the DNNs, a special focus is given to the advanced DL models such as CapsNets and SNNs. This chapter provides an overview of the DL systems’ functionality and the challenges related to the specific goals. Section 2.1 discusses a general background on DNNs followed by the latest related work on hardware accelerators for DNNs and optimizations for energy efficiency. Section 2.2 provides an overview of the CapsNets, with a special focus on the differences compared to the traditional DNNs. Section 2.3 discusses the background information on SNNs, neuromorphic architectures, and event-based vision. Section 2.4 discusses the DL vulnerability threats and the notable related works optimizing security and resiliency.