Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Discovery Grants Program - Individual

École Polytechnique de Montréal

$41,000.00

Computational methods and frameworks for unsupervised machine learning

EF 2024-25

Résumé
Data is everywhere in all activity sectors. Many organizations have faced the problem of having a lot of data with poor utility. Machine learning (ML) models have been extensively used in practice to extract knowledge from data. Whether it is to predict global warmth effects or to help detecting fake news, this quickly developing field often brings promising results. ML models can be categorized under two paradigms: (i) supervised learning, where learning is assisted by ground-truth data observations whose classes are known, denoted labelled data, and (ii) unsupervised learning, where no background knowledge is available a priori, and whose objective is generally set to capture the best underlying description of the data. Although supervised learning techniques can achieve minimal prediction errors and are successfully applied in numerous data analysis tasks, labelled data is often scarce and expensive to obtain, specially considering the velocity in which data has been generated in contemporary applications. The long-term objective of this project is to contribute to data mining practice associated with unsupervised learning tasks, by providing computational tools with proved effectiveness and demonstrated efficiency. The findings of this research program are expected to foster data-driven decision making for the Internet of Things (IoT) era in which huge amounts of unlabeled data will be generated in real-time from the most varied sources and devices (e.g. vehicles, sensors, home appliances, etc.) on an unprecedent scale. This ultimate goal will be supported by the accomplishment of the following three short-term objectives: A. Development of new optimization algorithms for clustering problems with guaranteed global optimality; B. Investigate the use of pairwise information for data embeddings; C. Design and implementation of a framework using evolving embeddings for data streams clustering and novelty detection. The proposed methods will combine combinatorial optimization techniques, by casting clustering problems and constraints as mathematical optimization problems; deep learning methods for yielding powerful data embeddings; and state-of-the art mining algorithms for anomaly detection and novelty detection in data streams. Benchmark and real-world instances will be used to validate our conclusions. The software and algorithms developed in this research program are expected to constitute new benchmarks to the field, allowing to approach larger datasets thereby leading to new insights in commerce, industry and academia. Moreover, given the high demand for data science/ML expertise in Canada, this research program will help to form and train specialists who will be able to master the scientific and technological issues emerging from modern data mining applications.
Programme
Discovery Grants Program - Individual
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
RGPIN-2023-04466
Name nom
Aloise, Daniel
Department département
Génie informatique et génie logiciel
Organization id
64
Institution établissement
École Polytechnique de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2023
Award amount
41000
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
1507
Committee name en
Computer Science
Committee name fr
Sciences informatiques
Area of application code
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
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)
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
Computational methods and frameworks for unsupervised machine learning
Keywords
unsupervised learning; clustering; combinatorial optimization; data embeddings; novelty detection
Application summary
Data is everywhere in all activity sectors. Many organizations have faced the problem of having a lot of data with poor utility. Machine learning (ML) models have been extensively used in practice to extract knowledge from data. Whether it is to predict global warmth effects or to help detecting fake news, this quickly developing field often brings promising results. ML models can be categorized under two paradigms: (i) supervised learning, where learning is assisted by ground-truth data observations whose classes are known, denoted labelled data, and (ii) unsupervised learning, where no background knowledge is available a priori, and whose objective is generally set to capture the best underlying description of the data. Although supervised learning techniques can achieve minimal prediction errors and are successfully applied in numerous data analysis tasks, labelled data is often scarce and expensive to obtain, specially considering the velocity in which data has been generated in contemporary applications. The long-term objective of this project is to contribute to data mining practice associated with unsupervised learning tasks, by providing computational tools with proved effectiveness and demonstrated efficiency. The findings of this research program are expected to foster data-driven decision making for the Internet of Things (IoT) era in which huge amounts of unlabeled data will be generated in real-time from the most varied sources and devices (e.g. vehicles, sensors, home appliances, etc.) on an unprecedent scale. This ultimate goal will be supported by the accomplishment of the following three short-term objectives: A. Development of new optimization algorithms for clustering problems with guaranteed global optimality; B. Investigate the use of pairwise information for data embeddings; C. Design and implementation of a framework using evolving embeddings for data streams clustering and novelty detection. The proposed methods will combine combinatorial optimization techniques, by casting clustering problems and constraints as mathematical optimization problems; deep learning methods for yielding powerful data embeddings; and state-of-the art mining algorithms for anomaly detection and novelty detection in data streams. Benchmark and real-world instances will be used to validate our conclusions. The software and algorithms developed in this research program are expected to constitute new benchmarks to the field, allowing to approach larger datasets thereby leading to new insights in commerce, industry and academia. Moreover, given the high demand for data science/ML expertise in Canada, this research program will help to form and train specialists who will be able to master the scientific and technological issues emerging from modern data mining applications.