Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Q* Approximation for Batch Reinforcement Learning: A Theoretical Comparison (arxiv.org)
21 points by ericzawo on Nov 23, 2023 | hide | past | favorite | 3 comments


Related from two days ago: https://www.reuters.com/technology/sam-altmans-ouster-openai...

"Some at OpenAI believe Q* (pronounced Q-Star) could be a breakthrough in the startup's search for what's known as artificial general intelligence" and "wrote a letter to the board of directors warning [it] could threaten humanity"


It seems like people are really stretching here. Q-learning has been a thing for a while now. And in optimization of X, Q, etc. the star is just used to mean the optimal value.


I agree. People are digging up papers with “Q*” in the title without understanding what they’re talking about. I think it’s silly to read into the OpenAI statement but their “Q*” algorithm probably means something pretty specific to them that would be very hard for us to guess from the name “Q*”




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: