•Updating
sequence: start at random state and act until it reaches absorbing goal
state
•For the
first updating sequence and our grid world example, how many weights get updated from the
first sequence?
•What
could we do if we kept the whole sequence in memory?
