Stellar inference speed via AutoNAS
Practical AI
00:00
Optimizing Neural Architectures for Edge Computing
This chapter explores the design of neural architectures tailored for specific hardware, introducing a proprietary algorithm, AutoNAC, for efficient neural architecture search. It illustrates the optimization of an image classification model, ResNet 50, while addressing the challenges and strategies of deploying optimized models in edge computing environments. The discussion emphasizes the relationship between latency and accuracy, and the importance of continuous innovation in neural architecture amidst evolving deep learning advancements.
Play episode from 17:04
Transcript
