Sergey Bartunov: One-shot generative modelling

There are many existing approaches to generative modelling which appeared recently such as variational autoencoders or adversarial networks. However, most state of the art models are able to produce good results (in terms of visual quality or likelihood) only after extensive training on large datasets. This talk will cover an emerging trend in generative modelling which is often referred to as one-shot learning, i.e. the ability to learn only on several training examples. In addition, a draft of the new mod