Transfer learning
DeepMind just published a mind blowing paper: PathNet.
Théo Szymkowiak

DM collects discreet data do efficiently, it doesn’t need to utilize superstitional chunking (totally just made that up, not sure the correct term) like humans do to summarize a concept enough to transfer learning.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.