Cambridge International AS & A Level Computer Science
Chapter 18 answers
What you should already know
1 Artificial Intelligence
is a machine which carries out a task that requires some degree of ‘human’ intelligence
duplicates human tasks requiring decision-making and problem-solving skills.
2 AI pros and cons
technologies such as autonomous vehicles
artificial limb technology helping the disabled
improvements in medical areas
job losses in many areas
dependence by humans on the technology
loss of skills since taken over by AI.
3 Examples of the use of AI
robotics
predictions such as climate change
performing delicate operations
bespoke cancer treatments
drones in bomb disposal, welding, nuclear incidents, and so on.
Activity 18A
Shortest route: A B C F I
Activity 18B
1 ‘A’ has two nodes ‘B’ and ‘F’
F(B) = 6 + 8 = 14, F(F) = 3 + 6 = 9
since F(F) < F(B) then ‘F’ is the next start node
Cambridge International AS & A Level Computer Science 1
© Helen Williams and David Watson 2020
Cambridge International AS & A Level Computer Science Answers
Path: A → F
‘F’ has two nodes ‘G’ and ‘H’
F(G) = ((3 + 1) + 5) = 9, F(H) = ((3 + 7) + 3) = 13
since F(G) < F(H) then ‘G’ is the next start node
Path: A → F → G
‘G’ has one node, ‘I’
so F(I) = 7 + 1 = 8
Path: A → F → G → I
‘I’ has three nodes ‘E’, ‘H’ and ‘J’
F(E) = ((3 + 1 + 3 + 5) + 3) = 15
F(H) = ((3 + 1 + 3 + 2) + 3) = 12
F(J) = ((3 + 1 + 3 + 3) + 0) = 10
F(J) has the smallest value …
… therefore shortest route is: A → F → G → I → J
2
Cambridge International AS & A Level Computer Science 2
© Helen Williams and David Watson 2020
Cambridge International AS & A Level Computer Science Answers
3 a)
b) shortest route: E B C D G
4 a) i)
ii) route order: A B C E G J
b) Time = Distance/Speed
T1 (A to J directly) = 56/95 = 35.4 minutes
T2 (A, B, C, E, G, J) = 50/80 = 37.5 minutes
direct route (dual carriageway) is the quickest
Cambridge International AS & A Level Computer Science 3
© Helen Williams and David Watson 2020
Cambridge International AS & A Level Computer Science Answers
Activity 18C
1 a) Narrow AI – machine which is superior to a human when doing a specific task.
General AI – machine which is similar in performance to a human in any intellectual task.
Strong AI – machine which is superior to a human in many tasks.
b) Reward and punishment
The machine learns from good results and bad results to help improve the
performance/produce optimal outcomes.
For example, search engines:
The number of hits shown on the first page (matching search criteria) is very high
indicates a good result.
If user needs to look at page 2, page 3, … then this is seen as a bad result and the search
engine would need to learn from these two outcomes to improve future performance.
c) Artificial neural networks:
Are artificial networks based on the biology of the human brain in its inter-connections
between neurons.
Makes use of input, output and hidden layers.
Has hidden layers are where the inputs are processed to produce output(s).
For example, the following diagram shows an artificial neural network (with two hidden
layers):
2 a) i) Supervised learning
are systems able to predict future outcomes based on past data.
uses labelled data
uses both inputs and outputs to train the model.
ii) Unsupervised learning
are systems able to identify hidden patterns based on input data provided
systems are not trained on ‘right’ answers.
iii) Reinforcement learning
are systems given no training …
… but learns on the basis of ‘reward and punishment’ when performing an action to
determine which action yields optimal outcomes.
Cambridge International AS & A Level Computer Science 4
© Helen Williams and David Watson 2020
Cambridge International AS & A Level Computer Science Answers
iv) Active learning
algorithms can interactively query source data to reach the desired output
makes use of labelled data and unlabelled data
makes use mainly of unlabelled data (based on cost reasons).
b) i) Back propagation
initial system nodes are given ‘weights’
actual output compared to expected outputs
if results not satisfactory then ‘weights’ are changed
‘weights’ updated until no errors or results are within acceptable limits.
ii) There are two types of back propagation: static and recurrent.
static maps static inputs to a static output
mapping is instantaneous in static but this is not the case with recurrent
training a network/model is more difficult with recurrent than with static
with recurrent, activation is fed forward until a fixed value is achieved
3 a) i) Supervised learning can decide if emails are junk emails.
ii) In marketing, unsupervised learning can look for patterns in groups of individuals.
iii) Reinforcement learning is used in gaming and robotics.
iv) Semi-supervised (active) learning is used in the classification of web pages.
b) i) artificial intelligence
ii) machine learning
iii) deep learning
iv) unlabelled (data)
v) web crawler (spider bot)
vi) chatbot
vii) back propagation (of errors)
viii) regression
ix) labelled (data)
x) punishment and reward
End of chapter questions
1 a) i) C
ii) D
iii) E
Cambridge International AS & A Level Computer Science 5
© Helen Williams and David Watson 2020
Cambridge International AS & A Level Computer Science Answers
b)
2 a) i) Reinforcement learning
given no training
learns using ‘reward and punishment’
uses optimisation techniques.
ii) Supervised learning
requires both input and output so model can be trained
once trained, uses labelled data and results compared to what they should be
uses regression and classification analysis.
b) i) Chat bots
interact through instant messaging, artificially replicating patterns of human interaction
uses machine learning.
ii) Search engines
measures success of search based on whether website/web page comes up as first page of
search
if not on first page, this is seen as a failure and system adapts to improve the search
process.
Cambridge International AS & A Level Computer Science 6
© Helen Williams and David Watson 2020
Cambridge International AS & A Level Computer Science Answers
iii) Photographic enhancement
some of the latest smartphones now use deep learning to give DSLR quality to the
photographs taken by the smartphone
the technology was developed by first of all taking the same photos using a smartphone
and then using a DSLR camera
the deep learning system was then trained by comparing the two photographs
then a large number of photographs already taken by a DSLR camera (but not by the
smartphone) were used to test the model.
3 random weighting
actual output
expected output
error gradients
minimised/removed
removed/minimised
back propagation
static/recurrent
recurrent/static
regression
4 a) Dijkstra’s algorithm will follow the path which is currently the shortest path, but pays no
attention to which direction we’re going in
A* builds on some idea of direction of the end point by adding an extra heuristic (H) value
which how far we have to go to reach the destination
Cambridge International AS & A Level Computer Science 7
© Helen Williams and David Watson 2020
Cambridge International AS & A Level Computer Science Answers
b) i)
ii) Route: C A S H N Q
5
Cambridge International AS & A Level Computer Science 8
© Helen Williams and David Watson 2020
Cambridge International AS & A Level Computer Science Answers
6) GPS will use A* algorithm:
to reset heuristic values based on possible alternative route(s)
re-calculate the shortest path/route based on f(n) = g(n) + h(n)
disregard any routes that take in nodes M-J, M-A-D-G-F
7 a) i)
a) ii) route: A B F G K L
b) A → D = 30 min}
D → H = 12 min} total = 62 mins
H → L = 20 min}
A → I = 20 min }
I → J = 14 min } total = 69 mins
J → K = 20 min }
K → L = 15 min}
Therefore reduced time (maximum) = 71 − 62 = 9 minutes
Cambridge International AS & A Level Computer Science 9
© Helen Williams and David Watson 2020
