Skip to main content
Workshops & seminars, Conferences & lectures

Seminar: Multimodal reasoning and clinical decision support


Date & time
Wednesday, April 5, 2023
10 a.m. – 12 p.m.
Speaker(s)

Samira Ebrahimi Kahou

Cost

This event is free

Organization

Department of Computer Science and Software Engineering

Contact

Weiyi Shang

Where

ER Building
2155 Guy St.
Room 1072

Wheel chair accessible

Yes

Abstract

    With recent advances in machine learning, the focus of AI research is shifting toward complex real-world applications. In these applications, it is often necessary to reason about data from diverse modalities, such as vision, text, or numerical data. In this talk, I am first going to present some of my early work in multimodal learning and reasoning. I will describe important challenges in multimodal representation learning, including short-cut learning and long-term credit assignment, with a focus on how they impact applications in clinical decision-making. Subsequently, I will present recent fundamental and applied research projects of my group in this problem space. Our offline reinforcement learning framework for treatment modeling enables the integration of several modalities and is amenable to expert guidance in the task of drug dosage recommendation. Our method for uncertainty-based intervention selection is designed to make concept bottleneck models, a class of models that enable interpretability and interventions, more robust in the presence of short-cuts. I will conclude the talk with an outlook on future work and how we intend to integrate these solutions into clinical decision support systems.

Bio

Samira Ebrahimi Kahou is an Associate Professor at École de technologie supérieure (ÉTS) in the Department of Software Engineering and Information Technology and an Adjunct Professor at McGill University in the School of Computer Science. She is a member of the Quebéc AI Institute (Mila) and of the Research Regroupment in Artificial Intelligence Applied to Critically Ill Children at CHU Sainte-Justine. She is also a Canada CIFAR AI Chair. Samira received her Ph.D. in Computer Engineering from Polytechnique Montréal/Mila with an award for the best thesis in the department. She graduated with honors in her M.Sc. in Computer Systems and Networks at the Kyiv Polytechnic Institute (KPI). Before joining ÉTS, Samira worked as a Postdoctoral Fellow at the McGill School of Computer Science and as a Researcher at Microsoft Research Montréal.

    Samira and her group focus on improving representation learning for sequential decision-making and applied machine learning. As a member of the regroupment, she works on applications of machine learning for patient monitoring and drug dosage adjustment to support clinical decision-making in ICUs. Samira’s work has been published in top-tier venues, such as NeurIPS, ICLR, ICML, ICCV, CVPR, and CoRL.

Back to top

© Concordia University