Hello. I am a computer science researcher specializing in machine learning and AI. I completed a postdoc at Carnegie Mellon University, and I received a PhD in Computer Science from the University of Texas at Austin.
Contact me at jonathanwilliammugan@gmail.com or on Twitter at @jmugan.
My research focuses on the goal of making the squishy reality of our everyday world available to computation.
I'm working on giving robots the ability to acquire a grounded understanding of their environment. See my recent article in The Gradient for a description of my current thinking. My PhD thesis resulted in an algorithm that enabled a robot to autonomously build models of its world using self-supervised learning. As a postdoc at Carnegie Mellon University, I explored how computers can use both context and feedback to learn the privacy preferences of users of mobile devices.I talk, write, and give tutorials on artificial intelligence and machine learning. Specifically, deep learning and natural language processing. I also do independent consulting, where I am currently focused on applying large language models to practical problems.
The Curiosity Cycle: Preparing Your Child for the Ongoing Technological Explosion
The Curiosity Cycle helps you prepare your children to be lifelong learners. Technology is rapidly changing our world, and those with curiosity will have the opportunity to thrive.
Learn More!Grammar checking using deep learning
We trained large language models (LLMs) for grammar checking. It turns out we were too early. Our models needed to be 10,000 times bigger.
Learn More!Upload your personality to the cloud as a Twitter cyborg.
Happy Cyborg is a web application that learns your personality and acts as you would on Twitter. It starts conversations, follows people based on your interests, and filters tweets to find the ones that you would enjoy reading.
Learn More!Fun with tracking, optical flow, image manipulation, and ray tracing.