IB Questionbank

User interface language: English | Español

Date	May 2019	Marks available	4	Reference code	19M.2.HL.TZ0.8
Level	HL	Paper	2	Time zone	no time zone
Command term	Describe	Question number	8	Adapted from	N/A

Question

Neural networks.

Genetic algorithms and neural networks are being used in a variety of scenarios. For example, a genetic algorithm may be used to organize timetables for trains whereas a neural network may be used to predict fluctuations between the exchange rates of different currencies.

Figure 1 shows an example of a neural network. It includes inputs, a hidden layer and outputs.

Figure 1: A neural network

Many toy companies are considering the use of machine learning using either supervised learning or unsupervised learning. MAGS, a large IT software company, has recently developed A Doll Called Alicia that allows children to interact with it.

A Doll Called Alicia uses machine learning to ensure the child can have the best possible communication with the doll.

Describe the difference between a genetic algorithm and a neural network.

[4]

Identify two ways in which the neural network could be modified that may improve its performance.

[2]

Describe the difference between supervised learning and unsupervised learning.

[4]

Explain why the machine learning capabilities of A Doll Called Alicia may lead to instances when the child and the doll cannot communicate effectively.

[4]

Companies such as MAGS are considering products that use unsupervised learning rather than supervised learning.

Explain the benefits of unsupervised learning in developing products such as A Doll Called Alicia.

[6]

Markscheme

Award [4 max].
A genetic algorithm works in the same way as an evolutionary process whereby it starts with a large population;
And uses an iterative process where the fitter solutions are selected and input into the next cycle until the exit criteria are satisfied;

Whereas neural networks attempt to mimic the process of the brain;
And can be used/trained to recognize patterns;

Award [2 max].
Increase the number of inputs;
Increase the number of hidden layers;

Award [4 max].
Supervised learning is when the outcome related to a given input is already known;
And so, the “learner” can recognize objects and name them based on the labels already given;

Whereas unsupervised learning is when no examples of outcome are given to help with the learning;
And so, the “learner” must deduce its own solutions e.g. classifying similar objects by colour or shape;

Award [4 max].
The language of the child may not have been programmed into the doll;
The child may not speak clearly;
The child’s language may not be sufficiently developed to apply syntax correctly;
The child may refer to something not in Alicia’s “recorded” content, or that Alicia has not previously “learnt”;

Award [6 max].
Unsupervised learning can be used for bridging the causal gap between input and output observations;
Instead of finding the causal pathway from inputs to outputs;
By building the model upwards from both sets of observations;
In the hope that the gap is easier to bridge in the higher levels of abstraction;
Possible to learn larger and more complex models;
e.g. the connection between two sets of observations;

Examiners report

Most candidates were able to provide a reasonable description of differences between a genetic algorithm and neural network.

The majority of the candidates answer this question correctly.

The majority of the candidates were able to provide a reasonable description of supervised and unsupervised learning. However, a few of them were unable to develop the responses beyond the generic points.

The majority of the candidates were unable to provide a reasonable response for this question. The responses were mostly limited to a few generic points.

Many candidates reasonably answered this question. However, the majority of the responses lacked variety in points specific to the question. The responses also lacked further explanation and were mostly limited to the identification of points and some description.

Syllabus sections

Option B: Modelling and simulation » B.4 Communication modelling and simulation

Show 29 related questions

Hide related questions

Option B: Modelling and simulation

Question

Markscheme

Examiners report

Syllabus sections

View options