人工智能代写 Foundations of Artificial Intelligence (CS5060)

人工智能 AI Artificial Intelligence

University of Aberdeen

Foundations of Artificial Intelligence (CS5060) 3rd Assessment (Various Topics)

Overview

This assessment will give you a chance to show your knowledge and understanding of topics studied in the course, as well as your problem-solving and communication skills.

Practicalities

The hand-in date for this assessment is

Lateness penalties:
• Handing in your submission on Tuesday will incur in a 10% penalty • Handing in up to a week late will lead to a 25% penalty.

Please notice:

  • This assessment will contribute towards 50% of your final mark for this course.
  • Plagiarism and collusion are serious issues and will not be tolerated. It is your responsibility

    to familiarise yourself with the University’s code of practice1. You must appropriately cite all

    material you have used in your work.

  • This is an individual assessment, not a team-based one.

    You should upload a single ZIP file in MyAberdeen, containing: • A document with all your answers as a PDF file
    • Any program, data and other files

    Scenario, Problem and Tasks

    1 Scenario

    A household has 4 major energy-consuming devices, listed below with their consumption: • A washing machine (500W)

    • A combined freezer/fridge (220W) • A dishwasher (1800W)
    • Electric underfloor heating (500W)

    The household is located in an urban area where electricity is charged depending on the time of the day, according to three bands:

    • Low band – between 11PM and 5:59AM, charged as 10p per kWh • Middle band – between 11AM and 4PM, charged as 15p per kWh • High band – all other times, charged as 20p per kWh

    It’s summer in Scotland and the temperature may range from 5oC– 25oC, with lower probabilities for low temperatures and higher probabilities for higher temperatures.

    Assume you have access to historical data recording when each device was turned on and off for a month.

    2 Problem

    Your task is to design an intelligent energy management system which would control the devices, turning them on or off during a week. Your solution should minimise costs by spreading the use of energy and using lower energy bands whenever possible.

09.00 BST, Monday 1st of October 2018

1 https://tinyurl.com/y92xgkq6

1

3 What you should provide

You should submit a technical report (up to 4,000 words) addressing the items below.

3.1 Choice of AI technique to represent/solve the problem

Choose any 2 of the following AI techniques studied in our course: • Search

• Logics (propositional or first-order logic) or production rules • Planning
• Decision trees
• Decision networks

To represent/model the problem and/or provide a solution.

As part of this item, you should

  • Provide a technical justification for your choice – explain why you think the techniques are

    adequate to represent/solve the problem.

  • Explain how you will use the techniques to represent and solve the problem. This explanation

    can be in the form of pseudo-code, or a list of steps you will go through to use/adapt the

    technique to represent and solve the problem.

  • If your solution takes into account the historic data on electricity usage (say, to predict future

    use) then you should explain what information should be recorded in the data. Important: Consider combining 2 techniques above, instead of using them independently.

3.2 Worked out example using techniques

In this item you should

  • Provide a worked out example of an instance of the problem and how you represent/model it

    using the techniques you chose in the previous item.

  • Explain how the instance of the problem is solved with each of the techniques you chose in the

    previous item.
    Your example should be non-trivial, that is, it should consider all devices and a whole week’s activities of the household.

3.3 A comparison of the techniques chosen

In this item you should provide a comparison between the two techniques chosen in terms of

• Guarantees (e.g., correctness, termination, etc.)
• Efficiency, accuracy, ease-of-use,
• Existence of available software/systems, and other technical issues.

3.4 Simple prototype and test cases

In this item you will implement a simple prototype to test out your ideas. Your prototype should make use of existing systems, instead of being implemented from scratch. For instance, depending on the techniques you have chosen, you may use the on-line PDDL editor from our tutorial2, or one of the

This item will contribute with up to 25% of the overall mark.

This item will contribute with up to 25% of the overall mark.

This item will contribute with up to 25% of the overall mark.

2 http://editor.planning.domains/

2

many existing implementations of decision trees3, or existing implementations of search algorithms4.

For this item you should submit:

  • The input (files) required to execute/run your prototype
  • Any code you have written and any piece of third party software your prototype needs in order

    to run – please indicate clearly what code is yours and what code is third-party

  • 2 test cases

    Your report should contain

This item will contribute with up to 25% of the overall mark.

4

• An explanation of how your prototype works
• Instructions to run your prototype (including how to use external Web sites or Web services) • The results of running your prototype and an explanation of how it solves the 2 test cases
• Issues you encountered (if any) and how you solved these
• A critical evaluation of the technologies you used

Marking scheme

Marks will be given based on
• Depth and breadth of knowledge

• Technical details of formalisation, worked out examples and pseudo-code • Communication skills (clear, technical contents and sound reasoning)
• Structure of document

3For instance https://www.geeksforgeeks.org/decision-tree-implementation-python/.
4For instance, https://tutorialedge.net/artificial-intelligence/breadth-first-search-java/.

3

Related Posts