Contextualize Me – The Case for Context in Reinforcement Learning

Carolin Benjamins, Theresa Eimer, Frederik Schubert et al.

Action editor: Adam White.

https://openreview.net/forum?id=Y42xVBQusn

#contextualize #contextual #reinforcement

Contextualize Me – The Case for Context in Reinforcement Learning

https://openreview.net/forum?id=Y42xVBQusn

#contextualize #contextual #reinforcement

The lack of #QTs may feel like a missing core feature, but consider: If I boost something without the ability to #contextualize or #comment, it'll look like I'm endorsing it. Which encourages me to boost #positive rather than negative messages.

To land here and demand, "We want the same plumbing and wiring we had in our old place! You know, the one over there, on fire, half submerged in shit?"

Could we maybe, I dunno, give it a month, and see if this is better somehow?