Explainable Deep Learning for Time Series in Healthcare

Description. This talk will cover key deep learning techniques for time series analysis, including Transformer-based models, Temporal Convolutional Networks (TCNs), and recurrent models such as LSTM and Mamba. We will explore their applications in healthcare and neuroscience, with a focus on model explainability and uncertainty quantification methods, such as conformal prediction. If time permits, we will review a recent study on seizure onset zone (SOZ) localization using intracranial electroencephalography (iEEG) signals from epilepsy patients with local model explainability.

Website: https://www.xaviermootoo.com/