Grant: Discovery Grants Program - Individual
University of Waterloo
$59,000.00
Explainable Data Systems
FY 2023-24
Summary
The modern world is powered by data. However, as the capabilities of data systems grow, so does their complexity. For example, language models that underpin conversational chatbots and computer vision models that identify objects in images need to learn millions, sometimes billions, of parameters from data. As a result of this complexity, the inner workings of data-intensive systems are not intuitive to end users and difficult for developers to troubleshoot. This creates critical barriers to adoption, especially in societal applications such as healthcare and law. In response, the European Union passed algorithmic explainability into law, through its General Data Protection Regulation (GDPR), and the United States government recently released an Artificial Intelligence (AI) Bill of Rights, one of whose five pillars is "Notice and Explanation: You should know when an automated system is being used and understand how and why it contributes to outcomes that impact you". To develop technologies that satisfy these requirements, a new research field has appeared: explainable AI. Within this field, the focus has been on explaining predictive models. For example, suppose a bank needs to explain a model that decides whether new loan applications should be approved. Here, an explanation may suggest that a client whose loan application was declined would have been approved if their annual income were 20 percent higher. As a contribution towards the emerging field of explainable artificial intelligence, the proposed research will produce new methods and software tools to advance the theory and practice of explainable data systems. The proposed solutions will confront the three Vs of big data: volume, velocity, and variety. To deal with very large datasets, efficient new methods will be proposed to find informative explanations quickly. In response to data velocity, explanation methods will be designed for systems that consume real-time data feeds produced by the Internet-of-Things and social media platforms. To address data variety, explanations will be developed to capture rich data semantics such as order, rankings, and entities and relationships extracted from text. The proposed tools will benefit society by enabling the use of data science in critical domains, such as healthcare, finance, education, climate action and law, where transparency and trust are required. Explanations will also benefit experts tasked with model auditing to ensure algorithmic correctness, regulatory compliance, data quality, fairness, and mitigation of bias. Trainees participating in this research will acquire sought-after skills in data systems, machine learning, and artificial intelligence. In all, this research will help position Canada as a global leader in responsible, inclusive, and trustworthy data science.
Program
Discovery Grants Program - Individual
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
RGPIN-2023-03670
Name nom
Golab, Lukasz
Department département
Management Science and Engineering
Organization id
33
Institution établissement
University of Waterloo
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2023
Competition year année de concours
2023
Award amount
59000
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
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
Artificial intelligence (computer vision, use2603)
Research subject en
Intelligence artificielle
Research subject fr
Intelligence artificielle (vision artificielle, utiliser 2603)
Application title
Explainable Data Systems
Keyword
data analytics; data science; data systems; explainability; machine learning
Application summary
The modern world is powered by data. However, as the capabilities of data systems grow, so does their complexity. For example, language models that underpin conversational chatbots and computer vision models that identify objects in images need to learn millions, sometimes billions, of parameters from data. As a result of this complexity, the inner workings of data-intensive systems are not intuitive to end users and difficult for developers to troubleshoot. This creates critical barriers to adoption, especially in societal applications such as healthcare and law. In response, the European Union passed algorithmic explainability into law, through its General Data Protection Regulation (GDPR), and the United States government recently released an Artificial Intelligence (AI) Bill of Rights, one of whose five pillars is "Notice and Explanation: You should know when an automated system is being used and understand how and why it contributes to outcomes that impact you". To develop technologies that satisfy these requirements, a new research field has appeared: explainable AI. Within this field, the focus has been on explaining predictive models. For example, suppose a bank needs to explain a model that decides whether new loan applications should be approved. Here, an explanation may suggest that a client whose loan application was declined would have been approved if their annual income were 20 percent higher. As a contribution towards the emerging field of explainable artificial intelligence, the proposed research will produce new methods and software tools to advance the theory and practice of explainable data systems. The proposed solutions will confront the three Vs of big data: volume, velocity, and variety. To deal with very large datasets, efficient new methods will be proposed to find informative explanations quickly. In response to data velocity, explanation methods will be designed for systems that consume real-time data feeds produced by the Internet-of-Things and social media platforms. To address data variety, explanations will be developed to capture rich data semantics such as order, rankings, and entities and relationships extracted from text. The proposed tools will benefit society by enabling the use of data science in critical domains, such as healthcare, finance, education, climate action and law, where transparency and trust are required. Explanations will also benefit experts tasked with model auditing to ensure algorithmic correctness, regulatory compliance, data quality, fairness, and mitigation of bias. Trainees participating in this research will acquire sought-after skills in data systems, machine learning, and artificial intelligence. In all, this research will help position Canada as a global leader in responsible, inclusive, and trustworthy data science.