The weight change rule for the ith neuron in an associative reward-penalty (ARP) network takes the form Awij = n[ (yi-f(
Posted: Wed May 11, 2022 6:11 am
The weight change rule for the ith neuron in an associative reward-penalty (ARP) network takes the form Awij = n[ (yi-f(a)r + 2(1-yi-f(a))(1 – t) ]x; (i) By considering all possible combinations of the binary output y; and the binary reinforcement r for this action, show that this rule implements an effective means to improve performance in an initially unknown and uncertain environment. [9 marks] (ii) What is the role of the parameter 1 in the learning process? Why is it not a good idea to set this parameter to zero? [4 marks)