Brain, cognition and machine learning
My main goal as a researcher is to develop new ideas, algorithms and mathematical models to extend the frontiers of science
and technology so as to improve the quality of life of humans and their environment. Mine is also a search for knowledge and a
desire to understand mind, cognition and rationality.
To this end, I conduct research in the following areas: Machine learning: Prediction and classification, sequential Monte Carlo and particle filtering, MCMC, variational inference, stochastic approximation, Bayesian statistics, optimization, probabilistic graphical models, structured relational stochastic models, active learning, online learning, unsupervised, semi-supervised and imitation learning. Cognitive Science and neural architectures: Cognition, sub-consciousness, sparse coding, Boltzmann machines, deep feature learning and invariance. Computer vision: Object recognition, image tracking and dynamic scene understanding. Robotics: Planning, navigation, sensing and actuation. Optimal control: Model predictive control, LQG, partially observed Markov decision processes (POMDPs) and reinforcement learning. Web-scale learning: Search engines, multimedia, web mining, social networks, collaborative filtering and recommender systems. Game theory: Sparse game representations and stochastic games.
Lately, I've been particularly interested in neural architectures and learning from web-scale datasets. The conjecture is that with a sound theory of intelligence, the right architectures and enough data, we might discover very simple (yet powerful) algorithms for perception, motor control and probabilistic reasoning.
To this end, I conduct research in the following areas: Machine learning: Prediction and classification, sequential Monte Carlo and particle filtering, MCMC, variational inference, stochastic approximation, Bayesian statistics, optimization, probabilistic graphical models, structured relational stochastic models, active learning, online learning, unsupervised, semi-supervised and imitation learning. Cognitive Science and neural architectures: Cognition, sub-consciousness, sparse coding, Boltzmann machines, deep feature learning and invariance. Computer vision: Object recognition, image tracking and dynamic scene understanding. Robotics: Planning, navigation, sensing and actuation. Optimal control: Model predictive control, LQG, partially observed Markov decision processes (POMDPs) and reinforcement learning. Web-scale learning: Search engines, multimedia, web mining, social networks, collaborative filtering and recommender systems. Game theory: Sparse game representations and stochastic games.
Lately, I've been particularly interested in neural architectures and learning from web-scale datasets. The conjecture is that with a sound theory of intelligence, the right architectures and enough data, we might discover very simple (yet powerful) algorithms for perception, motor control and probabilistic reasoning.
NEWS, ETC :
-
A special moment with some of the people who share my passion and path
Sam, Max, David, Irina, Marina and Jeff. We shall all miss Sam deeply not only for his brilliancy, but also for his enthusiasm,
collegiality and friendliness. He was that kid that was super friendly to me when I attended nips for the first time back in 1997.
- Slides of the NIPS 2009 Tutorial on SMC:
- CIFAR slides, where
Bo Chen and I
introduced an invariant, distributed, convolutional RBM for video.
Given a localized gaze in this scene, the model hallucinates the rest of the scene.
The hallucination is used to plan where to gaze next in order to solve a variety of tasks: object confirmation (dynamic recognition),
object detection, and identification of scene properties.
- A
Georgia Straight article about the "semantic" web.
- Worio is a web discovery engine. It has pretty cool machine learning: e.g. fast, accurate
classifiers for many millions of webpages and millions of classes. More info at
Worio press

- I'm always looking for bright PhD candidates with strong and diverse first degrees
(physics, math, stats, neuroscience, engineering, comp sci). If you're one of these people, please
apply to UBC here