The personal website is relocated to: https://blp.super.site/

All of sudden, Sergey Levin’s paper Control as Inference makes a lot of sense to me. Reinforcement learning, in its essense, is estimation. The difference is that the true value spans cross time. So the inferencing is done with time.

I could watch swimming treadmill videos like this for hours... pic.twitter.com/fZtNyIWOvI— Kyle Sockwell (@kylesockwell) July 19, 2022 [Read More]

I think I had a fundamental misjudgement on things. Originally, I thought that the people in power now are the group of zeilist in the 1960s. This is a generation without much education and exposure to international norms. For them the formative experience is the culture revolution where the cutting part of human nature is accecurated. When the younger generation moves into leadership postions, issues would be resolved and the country would embrace international norms like any other countries in the world. But education and exposure will only be superfacial if the things that motivotes people are ego. The governmental... [Read More]

[Read More]