$($a$3)$ Consider a reinforcement learning agent $($say for learning TIC TAC TOE$)$ instead of paying against random opponent. the agent played against itself., with both sides learning $..$ Under what conditions will the learning happen?Would it be different policy for selecting moves than playing with a human expert?

B$)$Consider the following temporal differencing rule.

V$($Si$)<-$V$($Si$)+$a$[$V$($Si$+1)-$V$($Si$)]$

How do we choose appropriate values for a to encourage convergence.Explain with all the necessary details.

Write an interesting problem$($in not more than $5$ sentence.where Reinforcement Learning could be used to solve the problem related to remote sensing$.$