AI & Inference

AI is about decision making, and an important component is inferring probabilities, which relies on probabilistic inference. CAIDA researchers work on exact and approximate probabilistic inference, particularly advanced Monte Carlo methods and variational inference for graphical models, lifted inference for relational models.  These methods are applied to domains of biology, medicine, and geology. 


Context of Work Area:

Statistical Machine Learning has the potential to be applied to almost every other area of study, making it incredibly valuable.  By adding statistics to machine learning, the use of artificial intelligence becomes more accurate and targeted. 

When you combine AI with Inference, you are able to take all of your observed data and convert it into something that you understand and care about very efficiently, while also giving you a proper characterization of your uncertainty in that thing. Before the integration of AI, large amounts of data would have made it very difficult to do this, but now inference appears in almost every statistical analysis of data.

UBC has some incredible leaders and up-and-comers in this area, whose work is helping to revolutionize the world of AI. From improving inference algorithms to ensuring algorithms are rigorous, these UBC researchers are not only making incredible strides in their own regards, but they are also providing other researchers with the tools they need to increase the quantity and discernment of their own work; thereby revolutionizing a myriad of fields.


Shared Interests:

  • Computational Statistics
  • Bayesian analysis
  • Optimization-based inference
  • Inference algorithms; Variational inference
  • Sampling-based inference algorithms; Markov chain Monte Carlo
  • Fundamentals of inference



David Poole
Trevor Campbell
Alexandre Bouchard-Cote