Matthew Schlegel

A PhD student at the University of Alberta in Edmonton actively researching in reinforcement learning and machine learning.



General Value Function Networks.

Matthew Schlegel, Adam White, Andrew Patterson, Martha White. Arxiv, 2018.

Adapting kernel representations online using submodular maximization.

Matthew Schlegel, Yangchen Pan, Jiecao Chen, Martha White. International Conference on Machine Learning (ICML), 2017.

Stable predictive representations with general value functions for continual learning.

Matthew Schlegel, Adam White, Martha White. Continual Learning and Deep Networks workshop at the Neural Information Processing System Conference, 2017.

Research Interests

Reinforcement learning

In the monolithic field of artificial intelligence reinforcement learning is focused on learning from interaction to maximize the reward recieved from an environment. Influenced from the Pavlovian conditioning (AKA: classical conditioning) and other psychological models, learning through extended streams of experience based off of delayed reward signals has the most potential to lead us to a general model for artificial intelligence (at least in my opinion). I work on many aspects of the reinforcement learning problem, but currently am focused on the use of general value functions for a means of representation.

Deep Learning

Deep learning is an extremely powerful way of transforming data into a predictive model without knowing anything about the distribution the data is sampled from. While it does have many problems (variance vs bias problem, high data requirements, and long training/convergence times), the models have demonstrated potential. I focus on how to learn deep networks for reinforcement learning (specifically for predictions with policy evaluation) in an online fassion. I also am interested in thinking of multi-network systems and where to disrupt training signals in the back propagation update. The debate on whether our systems should follow the tabula rasa or innatism ideologies keeps me up at night.

Physics and Space

Physics played a huge role in my early academic life, and continues to be a source of fascination. Rocketry and space are a consistent focus of my interests but also enjoy following the work from CERN with the Large Hadron Collider.

I must study Politicks and War that my sons may have liberty to study Painting and Poetry Mathematicks and Philosophy. My sons ought to study Mathematicks and Philosophy, Geography, natural History, Naval Architecture, navigation, Commerce and Agriculture, in order to give their Children a right to study Painting, Poetry, Musick, Architecture, Statuary, Tapestry and Porcelaine.

John Adams, 1780