Real-world is messy and complex. In critical applications such as healthcare, understanding AI/ML model decisions and knowing its limitations can go a long way in building trust. But what should you choose -- inherent interpretability or explainability? Are these same/different? Interestingly, like other real-world problems this depends on the context, and there is no one-size-fits-all approach. Drawing examples from my research, I will deconstruct how identifying needs of the problem requires a close collaboration between AI/ML experts and domain practitioners to build trustworthy ML models for the real-world.
Speaker bio: Dr. Sirisha Rambhatla is an Assistant Professor in the Management Sciences Department, Faculty of Engineering at the University of Waterloo (UW) where she leads the Critical ML Lab. Her research focusses on developing reliable machine learning (ML) and artificial intelligence (AI) models for critical real-world decision making in surgery, transplantation and healthcare, misinformation, and intelligent manufacturing using time-series and spatiotemporal modeling, representation learning, and explainable AI. Her inter-disciplinary work spanning both theory and practice of ML, has been published at top ML venues such as NeurIPS, ICLR, KDD, IJCAI, AAAI, and clinical venues such as AMIA, Urology Clinics North America, Surgery, and American Association for the Study of Liver Diseases. Recipient of the 2021 Merit Award for Excellence in Postdoctoral Research at the University of Southern California across Science and Engineering, Dr. Rambhatla recieved her Ph.D. and Masters in Electrical Engineering from the University of Minnesota -- Twin Cities in 2019 and 2012, respectively, where she was the recipient of the E. Bruce Lee Memorial Fellowship.
Please register on our Eventbrite page for in-person attendance. Pizzas and non-alcoholic beverage will be provided to in-person attendees.