Visual Reinforcement Learning (RL) agents trained on limited views face significant challenges in generalizing their learned abilities to unseen views.

The variability of what "similar" means in this context, which transformations are allowed, and whether data points themselves or their feature representations

The key idea behind Bayesian meta-learning is empirical Bayes inference of hierarchical model.

For example, the déjà vu method shows that for certain representation learning models and training images, it is sometimes possible to correctly predict the foreground label given only the representation of the background - better than through dataset-level correlations.