Dialogue Systems

I’ve spent the last few years of my PhD working on dialogue systems. Our work is focused on developing Deep Learning models for a wealth of problem domains including technical support systems, debating systems and chit-chat-oriented systems. Please see the publications page for further details on my work on dialogue.

I’m also currently co-leading the University of Montreal team for Amazon’s Alexa Prize competition – a competition for which Amazon has awarded our research group 100.000 USD in order to participate.

Learning to Play Chess with Deep Reinforcement Learning

In 1997, after years of engineering and hand-crafting, IBM’s Deep Blue managed to beat Garry Kasparov in chess. But isn’t there a simpler way to build a strong computer chess player? Yes – of course – by using Deep Reinforcement Learning!

Predicting the Stock Market from Twitter

Can the stock market be predicted from what people say on Twitter? For certain stocks, perhaps yes!

Discovering Human Emotions Automatically

Machine learning applied to discover emotions from humans watching music videos.

Learning and Inference in Conditional Random Fields

My undergraduate thesis on a class of probabilistic models known as Conditional Random Fields. Also includes a software program for image denoising.

Graphical Models and Their Applications in Data Analysis

Research project which investigates a broad range of graphical models and their corresponding applications.

Project Osiris

A non-profit platform working with statistical and sociological models, which I served as project coordinator for. Published a number of articles related to the growing poverty in Denmark and its complex relationship to education, health and life-style.