About me
Hey there! My name is Todd Morrill.
I’m an incoming PhD student in Computer Science at Columbia University advised by Professor Richard Zemel. During my master’s in CS at Columbia, I was also advised by Professor Kathleen McKeown. I am broadly interested in machine learning and tend to do a lot of work in NLP but have a few sub-areas that I’m interested in, including:
- neuroscience and cognitive science inspired approaches to machine learning. In particular, I am interested in local learning rules (e.g., Hebbian learning) or in other words, alternatives to backpropagation. I’m also interested in how sparsity can help create more disentangled representations. I’ve spent some time working with Hyperdimensional Computing (HDC) frameworks too.
- mechanistic interpretability. I am interested in understanding the algorithms that models implement with an eye toward making guarantees about their behavior and improving their performance.
- about a million other topics related to projects I’m involved in or just things I’m thinking about (e.g., risk control, continual learning, etc.)
I currently live in New York City with my fiancé and our sheepadoodle, Ziggy. Outside of work, I like to hike, camp, ski, and cook. In a former life, I loved to study languages and still speak Mandarin Chinese well enough. I also worked in industry for 10 years as a machine learning engineer at PwC implementing machine learning systems for our clients.
All views expressed on this site are my own.
