TinyML stands for Tiny Machine Learning which is basically as simple as deploying all the complex models that you hear about everyday on super small, cheap and convenient devices for easy of operation. I was instrumental in designing this course which was taught at University of Pennsylvania by my mentor and advisor Prof. Rahul Mangharam. This course was taught to more than 50 students spanning over an year. The students ended up developing cool projects which became self sufficient startups like a Bee Dance predictor for monitoring bee health to a dedicated vision based anchor deployment system for the navy. This was one of the best experiences of my time at Penn and one that enabled me to embrace the power of ML and sensors to solve everyday meaningful tasks.
Check out the website - https://tinyml-readthedocs.readthedocs.io/en/latest/api.html
This module focuses on the basics of Machine learning and the background knowledge required to understand and build TinyML applications. It answers the following questions:
What is Machine Learning?
What is TinyML?
How to think of about a TinyML problem?
What do you mean by Loss function?
What is Gradient Descent?
What are neurons?
What is the difference between training and testing data?
What are CNNs?
How to work with images?
What is responsible AI?Applying TinyML to real world applications after buidling a fundamental was the key to introducing it. We worked on different applications of ML from noise filtering on different audios, to visually identifying people using cameras to a gesture controlled application.
The final module involved students doing real world projects on hardware devices to creating their own voice and gesture controlled robot. More cool videos and details on the website