1 paper · 1 filter
Yuxiao Wen, Yanjun Han, Zhengyuan Zhou
We consider contextual bandits with graph feedback, a class of interactive learning problems with richer structures than vanilla contextual bandits, where taking an action reveals…