[Wednesday] Shard Theory
Details
Important note: this week only, moved to Wednesday!
Getting here: Enter the lobby at 100 University Ave (right next to St Andrew subway station), and message Giles Edkins on the meetup app or call him on 647-823-4865 to be let up to room 6H.
Shard theory is a research direction for reinforcement learning (RL) agents (although some of its findings may also be applicable to other areas of machine learning).
Rather than modeling a RL policy as having a single goal to "maximize the reward function" or "optimize for some other related goal", shard theory models them as a collection of contextually-activated, behaviour-steering computations. In some cases, these shards can be modeled as negotiating with each other.
And it may apply to human psychology as well!
Ariel will give a presentation on this interesting research direction, some of the evidence to back it up as well as the implications.
We welcome a variety of backgrounds, opinions and experience levels.
Every week on Thursday
[Wednesday] Shard Theory