Neural Architecture Search for Tiny Devices - Optimizing Deep Neural Networks for Edge Platforms

Join a comprehensive talk exploring Neural Architecture Search (NAS) for edge computing devices, presented by TCS Research Senior Scientist Swarnava Dey. Learn about the evolution of NAS since 2016 and its crucial role in developing Deep Neural Networks (DNN) for edge platforms. Discover how NAS enables automated learning of compact neural architectures while meeting specific hardware constraints like low latency, minimal memory footprint, and reduced power consumption. Gain insights into the fundamental NAS process and its underlying mathematical model, empowering TinyML engineers to effectively modify existing NAS frameworks. Understand the current challenges in balancing speed and accuracy in model generation, and explore solutions for creating efficient, customized models for target architectures.