MCUNet: Tiny Neural Network Design for Microcontrollers - Lecture 11

Explore algorithm and system co-design for tiny neural network inference on microcontrollers in this lecture from MIT's course on TinyML and Efficient Deep Learning Computing. Dive into the world of TinyML, focusing on microcontroller-based neural networks, TinyNAS, and TinyEngine. Learn how to overcome challenges in deploying neural networks on resource-constrained devices like mobile and IoT devices. Gain insights into efficient machine learning techniques, including model compression, pruning, quantization, neural architecture search, and distillation. Discover strategies for efficient training, such as gradient compression and on-device transfer learning. Explore application-specific model optimization for videos, point clouds, and NLP, as well as efficient quantum machine learning. Get hands-on experience implementing deep learning applications on microcontrollers, mobile phones, and quantum machines through an open-ended design project focused on mobile AI.

Lecture 11 - MCUNet: Tiny Neural Network Design for Microcontrollers | MIT 6.S965