Paper 3 – May/Nov 2018

About paper 3

The paper is normally written on the same day as paper 2 and has a duration of 1 hour, with a maximum mark of 30, counting for 20% of the total subject grade. Every year, it is a based on a case study or scenario that changes.

Here is the case study for the May/November 2018 exams.

The topic is Self Driving Cars and Machine Learning. 

Here is an in-depth summary of all the key concepts in this case study.

Here is a student revision guide – Thanks to Lucas Gurney (a 45 alumni student!)

Key concepts

A. Key terms

At the back of the case study, a list of 23 key terms that you need to understand and be able to discuss as definitions. Paper 3 May 2018 Key words contains a BASIC description of each term. Please do not accept these as gospel – do your own research!

B. Convolution Neural Networks

One of the biggest concepts in this case study is convolution neural networks. It is a very abstract concept, so here are a few videos to introduce it. Again, a lot of individual research is needed for this.

Video 1: Computerphile (14min)

Video 2: Deep Lizard IA (9min)

Video 3: MatLab (5min)

C. Self driving cars

Clearly, having a good understanding of how self-driving cars work would help you to understand the major theme of the case study. Here are a few videos, but again, a lot of individual research is required for this.

Video 1: Waymo (4min)

Video 2: The Hub (10min)

Video: Tedx Talk (10min)


D. Possible 8/12 mark questions to practice on:

These questions will usually be assessed using the Question 3/4 marking bands grid.

Q1: Self-driving taxis in Levangerstadt could cause several unintended consequences for the local labour market. Evaluate this claim.

Q2: CNNs require a lot of processing. Give a reasoned explanation of what processing should be done by the car’s local processors and what could be accomplished remotely via a network connection.

Q3: Any vehicle that operates in the Levangerstadt realm would need a fusion of sensors. Discuss how sensors could contribute to a safer transport system than the conventional human-driven one.

Q4: One of the many hurdles the system has to overcome is that of the “Trolley problem”. In the context of automating public transport, how could this problem possibly be addressed?

Q5: Discuss the ethical issues arising from beta-testing the self-driving systems in the Levangerstadt area.

Q6: As a diverse range of vehicles could foreseably be used in Levangerstadt, the importance of protocols is clear. Discuss how vehicle-to-vehicle and vehicle-to-infrastructure protocols could be implemented to ensure a more efficient system for the town’s transport users.