Canada CIFAR AI Chair Golnoosh Farnadi, Mila researcher and Assistant Professor in the Department of Decision Sciences at HEC Montréal, named among this year’s recipients of the Google AI Research Scholar Program for her project, “Addressing Algorithmic Fairness in Decision-focused Deep Learning.”
Launched in March 2020, the Research Scholar Program supports early-career faculty researchers from universities all over the world who are doing impactful work in the field of AI.
Dr. Farnadi has dedicated much of her academic career to developing novel machine learning and AI models to tackle fairness and ethics in AI. Her recent work has mainly focused on addressing bias and algorithmic discrimination in decision-making models.
Fighting Algorithmic Discrimination and Bias
AI and machine learning have become essential components in various automated tools that drive modern society. Recently, deep learning methods have gained widespread deployment in decision-making domains that affect people’s lives, from sentencing decisions to medical diagnostics and loan lending. Yet, such models are known to disproportionately discriminate and marginalize certain demographic groups. Dr. Farnadi wants to fix these shortcomings by developing mathematical tools and algorithms for effective and efficient fairness-aware deep learning approaches in automated decision-making.
“This research project is ambitious. If successful, it will be significant in all fields where deep learning models are involved in decision-making. These areas include education, housing, law enforcement, healthcare, banking, as well as new application domains yet to be discovered,” explained Dr. Farnadi.
With support from the Research Scholar Program, her goal is to design efficient deep learning algorithms that provide fairness guarantees, propose novel algorithms that ensure fairness in decision-focused deep learning models, and demonstrate the framework’s applicability to various areas.
While her research project has immense potential for AI ethics and fairness, Dr. Farnadi recognizes the long road ahead: “This process will be time-consuming and will require a lot of collaborative effort between ML researchers and policymakers. In addition to minimizing discrimination in automated systems, by properly building algorithms, we have the power to erase longstanding biases in criminal justice systems, policing, and many other areas of society, and provide a brighter future.”