Technical - Partner Event
AI Seminar – Roberto Vega Romero
When
July 16, 2021 - July 16, 2021
12:00 PM – 1:00 PM MST
Where
Online
Online
Title: Probabilistic Labels for classification tasks in medical images.
Abstract: Deep learning approaches often require huge datasets to achieve good generalization. This complicates its use in tasks like image-based medical diagnosis, where the small training datasets are usually insufficient to learn appropriate data representations. To compensate for the scarcity in data, we propose to provide more information per training instance in the form of probabilistic labels, which encode medical expert knowledge. We observe gains of up to 22% in the accuracy of models trained with these labels, as compared with traditional approaches, in three classification tasks: diagnosis of hip dysplasia, fatty liver, and glaucoma. The outputs of models trained with probabilistic labels are calibrated, allowing the interpretation of its predictions as proper probabilities. We anticipate this approach will apply to other tasks where few training instances are available and expert knowledge can be encoded as probabilities.
Short bio: Roberto Vega is a PhD candidate at the University of Alberta working under the supervision of Russ Greiner. His research focuses on how to combine machine learning with medical expert knowledge to learn accurate predictive models. He also collaborates with the local startup MEDO.ai, where he works alongside their AI team to automatically analyze ultrasound images for the early detection of hip dysplasia and thyroid nodules.
Looking to build AI capacity? Need a speaker at your event?
