Edge AI

Course Description

Edge AI is a rapidly growing field that combines artificial intelligence (AI) with edge computing. This integration allows AI models to run directly on edge devices—such as IoT sensors, smartphones, cameras, and microcontrollers—reducing reliance on cloud computing while enabling real-time processing and decision-making. In this course, students will learn the fundamentals of Edge AI, along with techniques and tools for implementing efficient AI models using various model compression and optimization methods. They will also gain hands-on experience in designing, developing, and deploying Edge AI applications. Additionally, the course covers advanced topics such as federated learning, sustainable AI, and responsible AI practices, with a focus on their practical applications in deploying Edge AI systems. By the end of the course, students will have the skills to design, develop, and implement end-to-end Edge AI systems for real-world applications across various industries.

The topics covered in this course are:

Foundations of IoT, Accelerated Edge Computing, and Edge AI
AI/ML and Computer Vision Fundamentals
Tiny and Embedded Machine Learning
Edge AI Hardware and Accelerators
Edge AI Software Frameworks and Deployment Pipeline
Model Compression and Optimization (Pruning, Quantization, Distillation)
Federated and Distributed Learning for Edge Devices
Generative and Agentic AI on Edge
Sustainable and Energy-Efficient AI
Case Studies in Smart Cities, Agriculture, and Healthcare

Instructor

Pandarsamy Arjunan

Assistant Professor
Centre for Cyber Physical Systems
Indian Institute of Science.

Teaching Assistants

Jare Suyog Machhindra
Moin

Course Logistics

Prerequisites

This course focuses heavily on hands-on learning. Students will work with embedded systems and program them using MicroPython and Arduino C. They will also learn how to implement and run efficient machine learning (ML) models using tools like TensorFlow and LiteRT. Therefore, proficiency in Python and C programming is essential. Additionally, a basic understanding of AI/ML concepts, microcontrollers, and IoT systems is highly recommended but not mandatory.

Grading

Five programming assignments (10 x 5 = 50%)
Paper presentation and discussion (10%)
Course project (40%)

Lab session and Hardware Platforms

There will be lab sessions every week (~1.5 hours) where students will learn how to program embedded systems using MicoPython and implement efficient ML/DL models that can be directly deployed on microcontrollers and IoT Edge devices. The Arduino Tiny Machine Learning Kit will be the main development board used in most sessions and for completing coding assignments. In addition, the following hardware platforms will also be provided, especially for completing the course project.

Text books and References

Machine Learning Systems by Vijay Janapa Reddi, Harvard University.
TinyML: Machine Learning with TensorFlow Lite on Arduino and Ultra-Low-Power Microcontrollers (2020) by Pete Warden and Daniel Situnayake TinyML Cookbook (2022) by Gian Marco Iodice.
AI at the Edge: Solving Real-World Problems with Embedded Machine Learning (2022) by Daniel Situnayake and Jenny Plunkett.
Deep Learning on Microcontrollers: Learn how to develop embedded AI applications using TinyML (2023) by Atul Krishna Gupta and Dr. Siva Prasad Nandyala.
TinyML: Machine Learning with TensorFlow Lite on Arduino and Ultra-Low-Power Microcontrollers (2020) by Pete Warden and Daniel Situnayake.
Hands-on TinyML: Harness the power of Machine Learning on the edge devices (2023) by Rohan Banerjee.
Dive Into Deep Learning (2023), Aston Zhang, Mu Li, Alexander J. Smola, Zachary Lipton.
Research articles.