BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Agi Meta-Objective: a Derivation
DTSTART:20260506T121500
DTEND:20260506T140000
DTSTAMP:20260429T174025Z
UID:286fd8714e309c350e978e0de99c912d35878c687067000af86a0761
CATEGORIES:Conferences - Seminars
DESCRIPTION:Clément Hongler\nMachines long learned to master games such a
 s go and chess. But these games were not designed to make machines more co
 gnitively able. Do such games exist and can machines create them to self-i
 mprove?\nWe consider the problem of building an artificial general intelli
 gence by self-play on self-creared sequences of games\, and of the associa
 ted problem of automated relevant skill discovery. We provide a concrete s
 ubstrate where to formulate the problem and argue that if there exists a g
 reedy optimization algorithm that achieves this\, then the form of the met
 a-objective satisfies gauge symmetries that lead to an explicit formula.\n
 \n 
LOCATION:BSP 727 https://plan.epfl.ch/?room==BSP%20727
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR