How DID YOU GET INTO machine learning?
I'm not really a machine learning researcher. I'm really an artificial intelligence researcher. That is, the goal of my career is to help produce intelligent machines, whether they are learning or not. It just happens that at the moment many (most?) of the interesting questions in AI are learning questions.
I decided I wanted to be a Computer Scientist in high school, after I taught myself to program in C++ - very slowly, and very painfully! - and found the whole thing incredibly cool. So then it was a matter of choosing the right field to specialize in. AI seemed like it had the biggest open questions and with the most potential for progress.
So during my undergrad at Wits I read all the AI textbooks I could find, and I did not find any of them particularly plausible, because they were devoid of plausible methods for building complete agents (even if they used an agent-centric definition of AI). That was encouraging, because there was clearly lots of room for progress. :) So I chose to focus on the areas of AI where the agent, and taking action, was the central thing: robotics, reinforcement learning, and planning.
I decided I wanted to be a Computer Scientist in high school, after I taught myself to program in C++ - very slowly, and very painfully! - and found the whole thing incredibly cool. So then it was a matter of choosing the right field to specialize in. AI seemed like it had the biggest open questions and with the most potential for progress.
So during my undergrad at Wits I read all the AI textbooks I could find, and I did not find any of them particularly plausible, because they were devoid of plausible methods for building complete agents (even if they used an agent-centric definition of AI). That was encouraging, because there was clearly lots of room for progress. :) So I chose to focus on the areas of AI where the agent, and taking action, was the central thing: robotics, reinforcement learning, and planning.
WhAT WILL YOU Be teaching?
I'll talk about advanced RL topics, following on from the introductory RL lecture. Specifically, we'll cover function approximation, policy gradient, and hierarchical RL.
What advice would you give to those getting started in machine/deep learning?
Pick a target that really means something, and go for it. Don't just write papers without a plan or direction. If you're good at research, you could do it for a lifetime. Your life's work should mean something.