Title: Robust simultaneous registration and classification model
Authors: Jian Qing Shi - Southern Univesity of Science and Technology (China) [presenting]
Abstract: An extended $t$ process-based two-level model is discussed, which allows simultaneously classifying and aligning functional data and provides a robust approach against outliers (either disturbed curves or disturbed discrete observed points in some curves). We use a logistic regression model and a data registration model to align and model the data at the same time, and also allow the models to use both scalar and functional variables. The trained models are applied to classify new data via an iterative procedure. The performance of the model is illustrated on both simulated and real data.