The inference models and the generative models paired in variational autoencoders (V AEs) are commonly constructed with neural networks, i.e., encoding networks and decoding networks, respectively

Imitation learning methods seek to learn from an expert either through behavioral cloning (BC) for the policy or inverse reinforcement learning (IRL) for the reward. Such methods enable agents to learn complex tasks from humans that are difficult to capture with hand-designed reward functions.

Designing flexible probabilistic models over tree topologies is important for developing efficient phylogenetic inference methods.

Deep neural networks (DNNs) have recently emerged as a powerful paradigm for solving Markovian optimal stopping problems. However, a ready extension of DNN-based methods to non-Markovian settings requires significant state and parameter space expansion, manifesting the curse of dimensionality.

To address these issues, we introduce a novel Bayesian meta-learning approach called meta-variational dropout (MetaVD).

However, in practice, the environment is typically time-varying and nonstationary .