1 paper · 1 filter
Lapo Frati, Neil Traft, Jeff Clune +1
This work identifies a simple pre-training mechanism that leads to representations exhibiting better continual and transfer learning. This mechanism -- the repeated resetting of we…