代做Network | network代写 | Artificial | 人工智能 | 神经网络 | Artificial Intelligence and Neural Computing – COMP0024 Artificial Intelligence and Neural Computing

COMP0024 Artificial Intelligence and Neural Computing

代做Network | network代写 | Artificial – 这个题目属于一个network的代写任务, 涵盖了Network/network/Artificial等程序代做方面

network代做 代写network 计算机网络

Artificial Intelligence and Neural Computing, COMP0024 () , year (Intended for all cohorts)

Main Assessment Period 2022-
There is ONE (1) question in total.
Marks for each part of each question are indicated in square brackets[n]
Total marks for this exam:[50]

Marks for each part of each question are indicated in square brackets.

Neural Computing

  1. a. A neuron receives external inputs 2.1, -0.7, -1.1, and 4.5, with corresponding synap- tic weights 0.2, 0.3, -0.5, and -0.2.Assuming a threshold of 0.5 in each case, cal- culate the output of the neuron under the following circumstances(for all of the following, write down the analytic formulate used to compute the correspond- ing output): i. The neuron is linear (no output transform applied), [6 marks] ii. The neuron is represented by a McCullough-Pitts model, [4 marks] iii. The neuron uses a sigmoidal firing function; [4 marks]
[Question 1 cont. over page]

COMP0024 1 TURN OVER

b. It is desired to store the 2 binary patterns [0.0, 1.0], [1.0, 0.0] in a 2-node Hopfield
net. Using the version of the Hopfield parameter setting rule that calculates values
for the neuron thresholds as well for their weights:
i. Calculate an algebraic form for the Hopfield energy function H(x).
[6 marks]
ii. Draw a state transition diagram for the system, labelling all transitions with
their probabilities, and showing the energy levels of the system;
[12 marks]
iii. Comment on when content addressable memory (CAM) properties of the sys-
tem associated to a Hopfield   network might not be ideal. Then contextualise
them to the system in the previous question.
[6 marks]
c. A reinforcement task for a single 2-inputARPunit is defined by the truth table in
Table 1 whereqy(x)is the probability of the neuron receiving a reward for actiony
x 1 x 2 q 0 (x) q 1 (x)
0 0 0.3 0. 2
0 1 0.8 0. 6
1 0 0.3 0. 2
1 1 0.7 0. 7
Table 1: Truth Table
in input contextxand the neuron has the initial parameter vectorw= (0, 0 ,0). The
probability of encountering pattern (0,1) is equal to the probability of encountering
(0,0). The probability of encountering pattern (1,0) is equal to the probability of
encountering (0,0). The probability of encountering pattern (1,1) is 2 times the
probability of encountering (0,0).
i. Calculate the probabilities of encountering each of the four patterns.
[4 marks]
ii. What is the value of the initial performance measureMinit?
[4 marks]
[Question 1 cont. on next page]

COMP0024 2 CONTINUED

iii. What is the value of the maximal performance possibleMmax?
[4 marks]
[Total for Question 1: 50 marks]

COMP0024 3 END OF PAPER