Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Discovery Grants Program - Individual

University of Alberta

$53,500.00

Learning structure models from complex data: principles and algorithms

FY 2009-10

Summary
Almost all data is now digitally stored, with unprecedented computing power available to process it. These developments have created new opportunities to advance computer interpretation (e.g. natural language processing, computer perception) and intelligent data-analysis (e.g. bio-informatics) by analyzing the massive amounts of complex data stored in text, multimedia, and scientific repositories. My research addresses the fundamental challenge of synthesizing predictive models from data, focusing on the problem of learning structured predictors in the presence of latent variables and incomplete observations. These challenges are motivated by the fact that predictions in complex domains are not just simple class labels or scalar values, but are complex structured outputs---such as parse trees, scene labellings or graph labellings---that involve multiple outputs to be predicted in a coordinated fashion, often with intervening latent variables. The key problem is training complex predictors when some of the output or intervening latent variables are unobserved. To tackle these problems, I employ two key strategies. First, I exploit probability models, which specify optimal interpretations from noisy, incomplete and conflicting evidence. Thus, one aspect of my research focuses on developing new algorithms for learning and inference in structured probability models. Second, I formulate convex training principles for the implied learning problems. Convexity decouples parameter optimization from model quality: a poor result arises from poor model structure, not a poor local minimum. A further key insight is that supervised training can be extended to unsupervised or semi-supervised training by following a self-supervised approach, where missing data components are treated as auxiliary variables to be optimized (i.e. inferred) simultaneously with parameter optimization. A natural convex formulation of joint training can then be obtained by working with equivalence relations over missing components. This approach has led to fundamental advances in unsupervised and semi-supervised training, including state of the art extensions to support vector machines, maximum margin Markov networks, and expectation-maximization.
Program
Discovery Grants Program - Individual
Location
Alberta, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
217337-2008
Name nom
Schuurmans, Dale
Department département
Computing Science
Organization id
9
Institution établissement
University of Alberta
Province en
Alberta
Province fr
Alberta
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2009
Competition year année de concours
2008
Award amount
53500
Program id
RGPIN
Program nane en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
331
Committee name en
Computing and Information Sciences - B
Committee name fr
Informatique - B
Area of application code
802
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Information systems and technology
Area of application fr
Systèmes d'information et technologie de l'information
Research subject code
2800
Research subject group en
Artificial intelligence
Research subject group fr
Intelligence artificielle
Research subject en
Artificial intelligence (computer vision, use2603)
Research subject fr
Intelligence artificielle (vision artificielle, utiliser 2603)
Partie
2
Nb partie
5
Application title
Learning structure models from complex data: principles and algorithms
Keyword
NA
Application summary
Almost all data is now digitally stored, with unprecedented computing power available to process it. These developments have created new opportunities to advance computer interpretation (e.g. natural language processing, computer perception) and intelligent data-analysis (e.g. bio-informatics) by analyzing the massive amounts of complex data stored in text, multimedia, and scientific repositories. My research addresses the fundamental challenge of synthesizing predictive models from data, focusing on the problem of learning structured predictors in the presence of latent variables and incomplete observations. These challenges are motivated by the fact that predictions in complex domains are not just simple class labels or scalar values, but are complex structured outputs---such as parse trees, scene labellings or graph labellings---that involve multiple outputs to be predicted in a coordinated fashion, often with intervening latent variables. The key problem is training complex predictors when some of the output or intervening latent variables are unobserved. To tackle these problems, I employ two key strategies. First, I exploit probability models, which specify optimal interpretations from noisy, incomplete and conflicting evidence. Thus, one aspect of my research focuses on developing new algorithms for learning and inference in structured probability models. Second, I formulate convex training principles for the implied learning problems. Convexity decouples parameter optimization from model quality: a poor result arises from poor model structure, not a poor local minimum. A further key insight is that supervised training can be extended to unsupervised or semi-supervised training by following a self-supervised approach, where missing data components are treated as auxiliary variables to be optimized (i.e. inferred) simultaneously with parameter optimization. A natural convex formulation of joint training can then be obtained by working with equivalence relations over missing components. This approach has led to fundamental advances in unsupervised and semi-supervised training, including state of the art extensions to support vector machines, maximum margin Markov networks, and expectation-maximization.
University of Alberta - $53,500 Grant (2009-10) | Canada Spends