Artificial Intelligence

Please note

This document only provides information for the academic year selected and does not form part of the student contract

School:

School of Computing and Engineering

Credit Rating:

Level (including FHEQ):

H (FHEQ Level 6)

Graded or Non Graded:

Graded

Version Valid From:

2023-09-25

Module Leader:

Emmanuel Papadakis

Version Number

2024.02

Learning Methods

Guided Independent Study

Practical Classes and Demonstrations

Lecture

Synopsis

Can machines (in particular computers) be intelligent? And what does that mean precisely? These are the main questions that we try to answer in this module. We will explore how machines can achieve intelligent tasks in a variety of settings. In term one we consider settings with full observability and determinism, these are like laboratory conditions or puzzle games. In this setting, we will look at knowledge representation, problem solving, and planning. In term two the settings are relaxed, and we will study how to deal with the uncertainties that arise from this. In particular, we will see how to deal with opponents, with incomplete and/or uncertain information, and how intelligent agents can learn.

Learning Strategy

Learning is to be achieved by attendance at lectures, seminars, and by study of directed reading. Attendance at seminars/tutorials will reinforce the taught material and give the students practical experience of symbolic intelligent systems construction and in the design, construction and training of Bayesian and neural networks.

Outline Syllabus

Underlying principles/techniques of symbolic AI: knowledge representation, inference and search. Examples are logic-based representation, logic inference, heuristic search, such as greedy search and A* algorithms. Overview of the characteristic capabilities of intelligent agents: reasoning, planning, learning, pattern recognition etc. Platforms and programming languages for AI (such as satisfiability solvers, Prolog, … For more content click the Read More button below. Basic computational models of intelligent functions such as planning – plan generation algorithms, plan model representations, planning tools such as planning engines, domain model parsers etc; and learning - clustering, reinforcement learning, Bayesian learning.

Overview of the characteristic capabilities of intelligent agents: reasoning, planning, learning, pattern recognition etc. Platforms and programming languages for AI (such as satisfiability solvers, Prolog, defeasible reasoning).

Basic computational models of intelligent functions such as planning – plan generation algorithms, plan model representations, planning tools such as planning engines, domain model parsers etc; and learning - clustering, reinforcement learning, Bayesian learning.

Learning Outcomes

On successful completion of this module students will

Explain concepts underlying symbolic AI such as knowledge representation and automated reasoning, and fundamental AI processes such as heuristic search methods.

Describe some of the central techniques in specific AI areas e.g., knowledge representation and/or machine learning.

Explain some of the main sub-symbolic approaches used to implement intelligent systems, such as neural networks or Bayesian networks.

Construct and reason with knowledge representations within a range of symbolic AI formalisms, such as rules, action schemas, classical logic.

Configure, apply and critically evaluate machine learning methods, and appropriate tools and techniques, for implementing intelligent systems in application areas.

Formative Assessment

Assessment 1: Quizzes / polls

Assessment 2: Written practice exercise

Summative Assessment

Assessment 1: Written Assignment

Assessment 2: Written Assignment

Description:One report designed to assess the understanding and practical application of sub-symbolic and uncertain AI principles such as Bayesian networks and machine learning.

Assessment Weight:50

Assessment Length:2000 Words (Word Count)

Module Learning Outcomes:MLO2, MLO3, MLO5

Individual or Group:Individual

Final Assessment Component:Yes

Marked Anonymously:No

Eligible for Tutor Reassessment:No

Assessment Criteria

Task 1: The work should display a sound grasp of the principles and practice of knowledge representation, heuristic search and inference, and it should demonstrate the ability to use and critically evaluate some of the respective tools. At Pass level, students will demonstrate a solution that shows a basic understanding … For more content click the Read More button below. Task 2: The work should display a sound grasp of the principles and practice of uncertainty and sub-symbolic AI, and it should demonstrate the ability to use and critically evaluate some of the respective tools. At pass level, students will demonstrate a solution that shows a basic understanding Bayesian networks, neural networks, and learning techniques. At distinction level, students will demonstrate a solution showing a deep understanding of techniques of uncertainty, sub-symbolic AI and learning.

Task 1: The work should display a sound grasp of the principles and practice of knowledge representation, heuristic search and inference, and it should demonstrate the ability to use and critically evaluate some of the respective tools. At Pass level, students will demonstrate a solution that shows a basic understanding of search methods (breadth first, depth first, best first, or A*), a basic knowledge of the form and role of heuristics, and some understanding of the form and use of knowledge representation languages such as satisfiability solvers. At distinction level, students will demonstrate a solution that shows a penetrating understanding of search methods and knowledge representation languages, the ability to use and predict the performance of individual heuristics, and an excellent ability to understand and manipulate sentences in a knowledge representation language.

Task 2: The work should display a sound grasp of the principles and practice of uncertainty and sub-symbolic AI, and it should demonstrate the ability to use and critically evaluate some of the respective tools. At pass level, students will demonstrate a solution that shows a basic understanding Bayesian networks, neural networks, and learning techniques. At distinction level, students will demonstrate a solution showing a deep understanding of techniques of uncertainty, sub-symbolic AI and learning.

Course and Module Catalogue