Hello again AIRGers (what is our demonym anyway?),
ïGiven that I will be presenting an AI game challenge next week (Hanabi), I thought I would take a moment to share another challenge I saw at NeurIPS this last December. Perhaps someone else could present on this one?
The challenge is called TextWorld, and the general idea is to train an RL agent to play a text adventure. Google Zork if you're unfamiliar with the concept of text adventures. Anyway, text adventures present a different set of challenges for agents from other video game genres (e.g. natural language processing) and I could bloviate why I think it's cool, but I'll leave it there.
https://arxiv.org/abs/1806.11532
https://www.microsoft.com/en-us/research/project/textworld/
https://competitions.codalab.org/competitions/20865
https://github.com/Microsoft/TextWorld
Finn Kuusisto, PhD
Postdoctoral Fellow
Regenerative Biology Laboratory
Morgridge Institute for Research
www.morgridge.org
|
![[Computer Systems Lab]](http://www.cs.wisc.edu/pics/csl_logo.gif){kind=logo}