Jacob Miller did his graduate studies at the University of New Mexico, where he developed protocols for carrying out measurement-based quantum computation using topological phases of quantum matter. After graduation, he took a year off to travel and reconnect with friends and family, before making the jump into machine learning research as a postdoc at Mila (Université de Montréal). Jacob’s research explores the use of tensor networks for quantum-inspired machine learning, with a focus on applications to natural language processing.
I want to contribute to a more complete understanding of how quantum methods can improve on the solution to problems in machine learning. In the same way as neural networks were inspired by the human brain, the methods I work with (tensor networks) were inspired by quantum systems, and I’m really curious to see how this influences their ability to solve human-level problems.
Natural language understanding. Classical deep learning methods have proven amazingly successful at generating text of near-human quality, but they do so in a way that makes it hard to disentangle the various factors guiding this process. By contrast, the convenient mathematical layout of quantum models gives us a much more robust foundation for understanding the different layers of structure they have picked up while learning to reproduce human language.
I have very eclectic interests and love learning about new subjects, with the result that I often spot connections between things that seem completely unrelated to everyone else.
John H. Conway, the English mathematician who (among many other things) invented the Game of Life and the concept of “surreal numbers”. He had this playful style of research which led him to results that looked absolutely absurd on the surface, yet were actually the first signs of much deeper phenomena.
Don’t do something just because it’s what others expect of you.
I played (American) football in high school. I was really bad at it, but have so many fun memories of my time on the team.