Data Sciences Institute

by Sara Elhawash

The day-long event was filled with engaging lightning talks, poster sessions, discussions, networking activities, interactive panels, and a focus on data-driven solutions for social good. It was a celebration of the remarkable interest, participation, and collective impact achieved by the DSI community. Moreover, it provided a platform for the DSI community to showcase their work and cultivate connections with collaborators from academia, industry, and government. 

The day began with a captivating keynote address delivered by Dr. Manuel Garcia-Herranz, Data Principal Researcher, UNICEF.  He emphasized the transformative power of Frontier Data technologies and their potential to address pressing global challenges. He discussed how the diversity and volume of data are reshaping technology’s capabilities and changing the world in profound ways. 

Dr. Garcia-Herranz also highlighted the importance of addressing data inequalities, noting that data from less privileged regions is often lacking. He underscored the need to bridge the gap between data scientists and those responding to real-world emergencies. One of the key collaborations highlighted during the event was the partnership between DSI and UNICEF on the Summer Undergraduate Data Science (SUDS) program.   

Research Day featured a series of enlightening lightning talks under the theme Data for Social Good, moderated by Bree McEwan, Associate Director, University of Toronto Mississauga, Data Sciences Institute and featured Madeleine Bonsma-Fisher, Data Sciences Institute Postdoctoral Fellow, Professor Marie-Josee Fortin FRSC, Canada Research Chair in Spatial Ecology (Department of Ecology & Evolutionary Biology, Faculty of Arts & Science), Assistant Professor Zahra Shakeri, Dalla Lana School of Public Health, Assistant Professor Nidhi Subramanyam, (Department of Geography & Planning, Faculty of Arts & Science). The talks showcased research that demonstrated how data science can improve lives and enhance human experiences. Topics ranged from equitable prioritization of active transportation infrastructure in Canadian cities to the use of anonymized movement data to assess urban park usage and more. 

Another set of lightning talks, Methodologies in Novel Applications, moderated by Ethan Fosse (Associate Director, University of Toronto Scarborough, Data Sciences Institute) featured Assistant Professor Jessica Gronsbell (Department of Statistics, Faculty of Arts & Science), Professor Peter Marbach (Department of Computer Science, Faculty of Arts & Science), and Senior Scientist Babak Taati (KITE Toronto Rehabilitation Institute, University Health Network). Their talks delved into innovative methods applied to critical issues in health and social sciences. Topics included auditing fairness in health applications, uniting sociological theory with data science concepts, and using machine learning to assess fall risk. 

During the networking lunch, the DSI community had the opportunity to engage with the excellent work of graduate students and trainees who presented their research via posters. 

by Sara Elhawash

How can innovative data-driven approaches like reinforcement learning revolutionize risk management for financial institutions? 

Financial institutions constantly deal with the challenge of managing risks tied to factors like interest rates, stock prices, and more. These risks, often unpredictable in nature, add complexity to the financial landscape. Managing risk is simpler when dealing with straightforward assets like stocks, where risks are typically linked to the asset’s price. However, complexities arise when dealing with financial derivatives like options, where risks are shaped by intricate non-linear relationships and unpredictable market changes. 

Historically, the financial industry has relied on parametric models to understand financial variable dynamics. The Black-Scholes model, introduced in 1973 by Black, Scholes, and Merton, became renowned for its constant-volatility assumption.  

DSI members and University of Toronto Professors Sebastian Jaimungal, (Department of Statistical Sciences, Faculty of Arts & Science) and John Hull (Joseph L. Rotman School of Management), propose a new frontier in financial risk management. Their aim is to develop alternative methods for quantifying and managing risk within financial institutions, utilizing reinforcement learning—a data-driven approach. Their strategies prioritize robustness to model misspecification and dynamic time consistency. 

Professor Sebastian Jaimungal explains, “Thanks to the invaluable support from DSI, our team has achieved a significant milestone with the development of ‘FuNVol: A Multi-Asset Implied Volatility Market Simulator using Functional Principal Components and Neural SDEs. This research employs Legendre polynomials to represent the surface and employs neural stochastic differential equations, a form of stochastic evolution driven by neural networks, to capture its intricate dynamics. With DSI’s support, we’ve been able to delve deeper into understanding volatility surface dynamics and its implications for risk management.” 

With the support of a DSI Catalyst Grant, this collaborative research team is working to better understand how volatility surfaces change using generative models. Their research has significant implications for risk assessment, risk management, and portfolio valuation, primarily benefiting financial institutions. Understanding the various ways volatility surfaces can evolve promises to enhance portfolio hedging strategies for financial institutions.  

“The DSI Catalyst Grant program underscores our commitment to advancing the frontiers of data-driven research, and we are delighted to witness the significant progress it has facilitated,” says Gary Bader, Associate Director, Research and Software, Data Sciences Institute.  

John Hull’s group is looking at the same challenge using variational autoencoders (VAEs), a model where latent factors determine option price location and spread. 

What sets this research apart is the goal of incorporating these generative models as inputs into reinforcement learning algorithms. Their aim is to develop sophisticated strategies for mitigating risks within portfolios of financial options and toward more robust and effective risk mitigation strategies in an ever-evolving financial landscape. 

by Sara Elhawash

Amidst the whirlwind of rapid digital transformation sweeping across industries, we are shining a spotlight on the pivotal role that data science plays in building an effective workforce. Workforce strategies will be showcased at the Data Sciences Institute Research Day on September 27, 2023. 

The panel titled “Data Science for an Effective Workforce,” will feature data science leaders from the private sector, non-profit organizations and the government. The panel will include David Campbell, Assistant Director, Data Science Applications at the Bank of Canada; Yves Jaques, Chief, Frontier Data & Tech Unit at UNICEF; Ann Meyer, Director, BioInnovation Scientist Program at adMare BioInnovations; Mark Fiume, Co-Founder and CEO, DNA Stack, and Dana Ohab, Associate Partner, Digital & Emerging Technology at EY. Each of the panelists will bring a wealth of insight on the topic. The event presents an exciting opportunity to explore the synergy of data science and modern workforce development in a world that’s becoming more data driven. 

As industries become more complex and interconnected, the ability to harness and interpret data has become essential for making informed choices that drive growth, efficiency, and innovation. 

Yves Jaques, Chief of the Frontier Data and Tech Unit at UNICEF, extends the perspective: “Data science is borderless. It defies geographical constraints, knitting together a digitally connected workforce that is not bound by location. It gives us the possibility to bridge the digital divide by creating resilient networks that empower our national partners to scale and sustain local solutions with local talent, capitalizing on the collective intelligence of a global workforce.”   

“The Data Sciences Institute Research Day serves as a platform for delving into the intricate interplay of data science and workforce strategies. It’s an opportunity to explore how these two domains coalesce to define the future of work,” says Lisa Strug, Director of the Data Sciences Institute. 

As a multi-divisional, tri-campus, multidisciplinary hub for data science activity at the University of Toronto, DSI brings together researchers and trainees from across the University, its affiliated research institutes, industry and beyond to support data sciences research, innovation, collaboration, and training to translate promising ideas into real-world solutions and advance the data sciences, themselves.  

Research Day #DataSciencesDay, serves as a platform for this discussion. Attendees can expect to learn from these insights through the panel discussion, lightning talks from DSI researchers, poster sessions and the invaluable networking sessions that promise to enrich understanding.  

For those interested in joining in on the DSI Research Day and gaining new insights, the countdown has commenced. Register here to secure your spot. 

by Sara Elhawash

As generative AI like ChatGPT and Large Language Models become increasingly integrated into our daily lives, how can we strike a balance between harnessing their potential for innovation and ensuring responsible and ethical usage? 

Funded through the Emergent Data Sciences Program competition, University of Toronto Professors Syed Ishtiaque Ahmed (Department of Computer Science, Faculty of Arts & Science), Shurui Zhou (Edward Department of Electrical & Computer Engineering, Faculty of Applied Science & Engineering), Lisa Austin (Faculty of Law), Shion Guha and Anastasia Kuzminykh (Faculty of Information), are co-leading Toward a Fair and Inclusive Future of Work with ChatGPT. 

The program focuses on the responsible development and ethical implementation of generative AI. It aims to shed light on the societal implications of using ChatGPT, with a particular emphasis on its impact on diverse communities. By gaining a deeper understanding of the social and ethical aspects of generative AI, the program seeks to empower researchers and users to make informed decisions and employ responsible practices when utilizing these technologies. 

It will feature a series of talks, discussions, and participatory design sessions involving individuals from various backgrounds, including students, instructors, practitioners, academics, and artists.  

“We recognize the profound influence of generative AI technologies on diverse communities. Our program seeks to bridge the gap in evaluation frameworks and provide a platform for diverse voices to express their experiences and insights. By fostering inclusivity and promoting ethical considerations, we aim to empower users and researchers to navigate the responsible use of generative AI with confidence,” says professor Shurui Zhou 

Activities of the program include events that will provide a platform for diverse perspectives and experiences with ChatGPT, workshops and public-facing meetups to foster inclusivity and encourage open dialogue, with a focus on amplifying the voices of minority communities. Academic workshops will be co-located with major conferences, such as the Conference on Human Factors in Computing Systems (CHI), Computer-Supported Cooperative Work & Social Computing (CSCW), and Neural Information Processing Systems (NeurIPS), to disseminate research findings and engage with a wider audience. To ensure ongoing interdisciplinary discussions and knowledge sharing, the researchers will create a repository of videos, talks, and posts, hosted on the Data Sciences Institute’s website, related to the societal implications of generative AI. 

A course syllabus module to educate students about the ethical considerations surrounding generative AI will also be developed. One of the unique aspects of the program involves students at the University of Toronto engaging in year-long projects that incorporate the use of ChatGPT within their workflows. This practical experience will enable students to share their findings and lessons learned through a poster presentation, contributing to collaboration and knowledge exchange. 

As the program kick starts its activities, Professor Shurui is conducting an interview study to understand how large language models (LLMs), such as ChatGPT, might affect the practices of scientists and research software engineers to collaborate and develop software. To participate, visit our website here. 

Recipients of the DSI Emergent Data Sciences Program competition are funded for their programs which foster the development of innovative data science methodologies, deep connections with computation and applied disciplines, new training programs, collaboration, knowledge mobilization, and impact beyond the academy. 

“This Emerging Data Science program is driven by a shared mission to assess the societal implications of generative AI. Together, we aim to create a robust framework that promotes trust, accountability, and transparency in the AI ecosystem, ensuring that these technologies benefit all members of society,” says David Lie, Associate Director of Thematic Programming & Data Access at the Data Sciences Institute. 

by Sara Elhawash

In an era where data-driven insights fuel innovation and inform decisions, policymakers and stakeholders increasingly seek guidance in research from various areas such as criminal justice, health, and labour law. However, the wealth of data gathered to understand human behaviour can lead to misguided recommendations if not approached appropriately during the analysis phase. This challenge has inspired the question: How can we elevate the quality of data analysis to better inform decision-making? 

Funded through the Emergent Data Sciences Program competition, University of Toronto Professors Linbo Wang (Department of Statistical Sciences, Faculty of Arts & Science, University of Toronto Scarborough), Gustavo J. Bobonis (Department of Economics, Faculty of Arts & Science), Ismael Mourifié (Department of Economics, Faculty of Arts & Science), and Raji Jayaraman (Department of Economics, Faculty of Arts & Science), are co-leading Bringing Together Data Science and Causal Inference for Better Policy Recommendations.  

The program promotes cross-disciplinary exchange and collaboration among experts in data science, causal inference, and applied research. Its overarching mission is to influence the landscape of data sciences by advancing the state of the art in causal inference and its applications to real-world policy problems. The program aims to tackle key challenges in data science, including algorithmic fairness, bias from confounding variables, and the need for more robust statistical inference methods. 

The program aims to achieve this by creating an inclusive forum for discussions across diverse disciplines. Here, researchers will get to share their research questions, data limitations, and challenges related to causal methods. Experts in data sciences and causality will introduce new and existing methods, encouraging the pursuit of research goals. Applied researchers will present key limitations informed by practice, jointly addressing the barriers to using current methods in solving policy problems of our time. 

Featured activities include three workshops and a lecture series on causality. In these workshops, data scientists, causal inference experts, and empirical researchers collaborating with policymakers convene to present their work. The lecture series focuses on sharing the state of literature with a non-specialized audience. The first workshop, Forging a Path: Casual Inference and Data Science for Improved Policy, is scheduled for November 10-11.  

“Our collaborative effort will enable us to address pressing policy questions with a newfound depth, ensuring that data-driven decisions are rooted in robust causal understanding,” say Professors Ismael Mourifié and Linbo Wang. “We look forward to working alongside fellow experts to drive meaningful impact in both academia and policymaking.” 

Recipients of the DSI Emergent Data Sciences Program competition are funded for their programs, which foster the development of innovative data science methodologies, deep connections with computation and applied disciplines, new training programs, collaboration, knowledge mobilization, and impact beyond academia. 

“This Emerging Data Science program exemplifies DSI’s commitment to fostering collaboration and innovation in data science research. It reflects our dedication to addressing complex challenges at the intersection of data analysis and real-world policymaking. We are confident that this initiative will have an impact,” says David Lie, Associate Director of Thematic Programming & Data Access, Data Sciences Institute.