With the help of UMass Amherst students, Erik is working to give machines the ability to react intuitively to new situations and learn from the outcomes.
I've always been interested in human intelligence and human behavior, and also whether the manifestations of human intelligence could be put into a machine. Learning from example with computers is now really well-established. What I'm super excited about is something called unsupervised learning. If we really want things that can just go out in the world and learn by observing, we need to master unsupervised learning. The iPhone 10 has face recognition for security access. The hope is that eventually, you could pick up a smartphone device, and it would identify who picked it up, and then swap in settings that they wanted, all in a completely seamless way. Maybe you don't want it reading your emails out loud unless you're alone. And so it's going to scan the room, see whether you're alone, and then do things condition on that. Computer vision can ultimately be used to do all those things. But of course, it's not just identifying the context, but it's also learning to condition your behavior on those contexts. The UMass students are great. We publish at the top conferences. And the papers that we write, I think, are respected worldwide. Anybody with a Bachelor of Science or Bachelor of Arts in computer science from UMass right now is in high demand from all kinds of industries. When a student comes to UMass, there's no limits to what they can do. I'm Erik Learned-Miller. I'm a professor of computer science. I stand for basic research and deep understanding, and I stand for UMass.