Streaming Inference for Infinite Feature Models

摘要

Unsupervised learning from a continuous stream of data is arguably one of the most common and most challenging problems facing intelligent agents. One class of unsupervised models, collectively termed