Grant: Discovery Grants Program - Individual
McGill University
$44,000.00
Active Learning and Individualized Variable Selection
EF 2024-25
Résumé
This research proposal outlines a multifaceted approach aimed at enhancing the fields of active learning and individualized variable selection. The first objective focuses on active learning through data valuation, which addresses the challenge of efficient data annotation by identifying and selecting the most informative samples. We propose a novel data Shapley value approximation method to improve batch active learning, particularly in large-scale settings. Our approach introduces a model-free, cost-sensitive perspective by considering real labeling costs.
The second objective deals with individualized variable selection and sequential variable selection. In a clinical context, it addresses the challenge of optimizing predictor measurement for disease risk prediction and treatment recommendation. We aim to develop a dynamic prediction strategy that determines the optimal sequence for measuring predictors, maximizing model performance while minimizing feature acquisition costs.
This research has wide-reaching implications, benefiting high-dimensional statistics, variable selection, and statistical computing. Additionally, the proposed methods will be applied to the field of drug discovery, offering cost-effective ways to select compounds for testing, improve pharmacokinetic models, and enhance drug-target interaction understanding.
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data
Tel que publié
Application id
RGPIN-2024-06780
Name nom
Yang, Archer Yi
Department département
Department of Mathematics and Statistics
Organization id
61
Institution établissement
McGill University
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2024
Award amount
44000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1508
Committee name en
Mathematics and Statistics
Committee name fr
Mathématiques et statistique
Area of application code
504
Area of application group en
Manufacturing processes and products
Area of application group fr
Procédés et produits industriels
Area of application en
Human pharmaceuticals
Area of application fr
Produits pharmaceutiques administrables à l'homme
Research subject code
3004
Research subject group en
Statistics and probability
Research subject group fr
Statistique et probabilité
Research subject en
Applied statistics
Research subject fr
Statistique appliquée
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Active Learning and Individualized Variable Selection
Keywords
statistical machine learning; statistical computing; high--dimensional statistics; data integration; transfer learning; generalized estimating equation; fixed-point problem; variational inequality; structural missing; semi-supervised learning
Application summary
This research proposal outlines a multifaceted approach aimed at enhancing the fields of active learning and individualized variable selection. The first objective focuses on active learning through data valuation, which addresses the challenge of efficient data annotation by identifying and selecting the most informative samples. We propose a novel data Shapley value approximation method to improve batch active learning, particularly in large-scale settings. Our approach introduces a model-free, cost-sensitive perspective by considering real labeling costs.
The second objective deals with individualized variable selection and sequential variable selection. In a clinical context, it addresses the challenge of optimizing predictor measurement for disease risk prediction and treatment recommendation. We aim to develop a dynamic prediction strategy that determines the optimal sequence for measuring predictors, maximizing model performance while minimizing feature acquisition costs.
This research has wide-reaching implications, benefiting high-dimensional statistics, variable selection, and statistical computing. Additionally, the proposed methods will be applied to the field of drug discovery, offering cost-effective ways to select compounds for testing, improve pharmacokinetic models, and enhance drug-target interaction understanding.