Evidential Turing Processes (ETP)

1. We define Total Calibration for the first time.
2. We introduce Complete Bayesian Models
3. We propose Evidential Turing Processes as a realization of Complete Bayesian Models
4. We show that Evidential Turing Process is the only model that can excel all three aspects of total calibration simultaneously

Continual Learning of Multi-modal Dynamics with External Memory (CDDP)

1. We introduce the first study on Continual Learning of Multi-modal Dynamical Systems
2. We introduce Competitive baseline Variational Continual Learning - Bayesian State-Space Model
3. We propose a novel solution which is Continual Dynamic Dirichlet Process (CDDP) based on external memory
4. We show that CDDP is beneficial for continual learning of multi-modal dynamics