Florent Krzakala
Home
Publications
Teaching
SPHINX group
Contact
CV
optimisation
Streaming Bayesian inference: Theoretical limits and mini-batch approximate message-passing
In statistical learning for real-world large-scale data problems, one must often resort to “streaming” algorithms which operate sequentially on small batches of data. In this work, we present an analysis of the information-theoretic limits of …
Cite
×