再补一句

document.tex

@@ -91,10 +91,13 @@ do not explicitly utilize exploration strategies
     We proved that the 2048 game is acyclic between non-absorbing states.
      This is the fundamental reason why explicit exploration cannot be 
      employed in the 2048 game.
+     Compared to explicit exploration strategies, 
+     backward learning, restart, and optimistic initialization
+     are more suitable for acyclic MDPs or MDPs with acyclic structures.        
 \end{abstract}
 
 \begin{IEEEkeywords}
-Acyclicity, 2048 game, ergodicity, backward learning.
+Acyclicity, 2048 game, ergodicity, exploration, backward learning.
 \end{IEEEkeywords}