Robotics

Module code: 7CCSMART
Credit level: 7
Credit value: 15
Semester:  Full-year 
Assessment:  coursework 

Aims
To teach you to read and understand research papers and research lectures on your own, and pursue a research topic.

Learning Outcomes
You should obtain a good understanding of a particular technical area at a level that goes substantially beyond the taughts MSc materia.  You should also learn to explore a research area, to identify the important issues and understand their connection with each other and to demonstrate your technical understanding by presenting the results to a scientifica audience.

Provisional Syllabus
In this optional module, you will study advance research literature preferably in an area that is related to the material taught in your programme.

7CCSMART is a first-term module, with work starting at the beginning of the first term, but with the assignment continuing into the second term. Lectures will cover research methods, report/paper writing, and presentation techniques. You must attend at least five selected research seminars during the first term and read several related research papers.

You must agree your choice of research topic with the module organiser, which must be on a topic of in the research area of one of the members of the Department. Towards the end of the first term, you must submit a first draft of a report (approximately 10-15 pages), which will be reviewed by the organiser and other students. The final draft of the report must be submitted in the second term. The submitted report forms the basis of the assessment.

Topics will include:
Doing Research
Writing
Writing Scientific Papers
Presenting Scientific Papers
References
Sources
Refereeing

Module code: 7CCSMAMS
Credit level: 7
Credit value: 15
Semester:  Semester 1 (autumn) 

Aims
To describe some techniques employed in the characterisation of agents and multi-agent systems. To provide a critical introduction to theories and methods regarding multi-agent computer systems and their component agents.

Learning Outcomes
On completion of the module, you will be expected to have acquired: A thorough, systematic understanding of key features of current theories and methods regarding multi-agent systems and their component agents; A sound appreciation of the conceptual issues involved in the characterisation of agents and their abilities; Knowledge of some of the main techniques employed in the formal characterisation of agents and multi-agent systems; An ability to critically evaluate current work in this field, and to evaluate the principal theories and methods.

Provisional Syllabus
Topics will be selected from:
Intelligent agents and their design
Knowledge in multi-agent systems
The Belief-Desire-Intention model of rational agents
Reactive and hybrid agent architectures
Agent Communication: KQML, FIPA
Auctions and Negotiations Game Theory Argumentation based Reasoning and Communication
Agent-based methodologies Applications

Module code: 7CCSMAIN
Credit level: 7
Semester:  Semester 1 (autumn) 

Over the last 40 years, Artificial Intelligence has developed into one of the core disciplines of computer science, combining symbolic reasoning (usually logic based), and optimised algorithms to provide solutions to complex and computationally difficult problems such as machine learning, visual recognition, natural language processing, planning and robotics. This module presents the main issues encountered in artificial intelligence and introduces approaches to deal with them.

Module code: 7CCSMDSM
Credit level: 7

This course aims to provide an overall understanding of the basic concepts and practical technology of distributed computing, an in-depth understanding of the considerations applied in designing software for distributed systems. It will give students knowledge of algorithmic and architectural techniques used to address these considerations, and provide means of assessing how well novel technologies are able to handle faults in distributed systems.

Module code: 7CCSMGPR
Credit level: 7

To provide the experience of working in a syndicate to design, implement and document a substantial software product.

Module code: 7CCSMOME
Credit level: 7
Credit value: 15
Semester:  Semester 2 (spring) 

Aims
To introduce various discrete optimisation problems, efficient algorithms for solving these problems, and general algorithmic techniques, which can be applied to a wide range of optimisation problems. The emphasis is put on network optimisation problems and on general optimisation techniques. To discuss applications of optimisation problems in communication systems, computer networks, manufacturing, scheduling, and resource allocation.

Learning Outcomes
On successfully completing this module you will be able to express computational problems from various application areas as (discrete) optimisation problems; will be familiar with commonly used algorithms and main algorithmic techniques for optimisation problems; will understand the principles underpinning the discussed algorithms; will be able to select an appropriate algorithm for a given optimisation problem or to develop a new algorithm based on a general algorithmic technique; will be able to analyse the running time of the developed algorithmic solutions.

Provisional Syllabus
Single-source shortest-paths problem:
Dijkstra's algorithm
The Bellman-Ford algorithm
Shortest paths in directed acyclic graphs
All-pairs shortest paths:
Johnson's algorithm
Network flow problems:
Maximum flows, Minimum-cost flows and Multicommodity flows, and their applications
Maximum matching problem and its applications to resource allocation problems
The Ford-Fulkerson method for the maximum-flow problem
The Successive-shortest-paths algorithm for the minimum-cost flow problem
Linear programming (LP):
Basic properties of LP problems
LP formulation of network flow problems
Integer programming
Computationally hard optimisation problems:
Polynomial-time problems and NP problems
NP-hard optimisation problems
Optimisation techniques for NP-hard problems:
Branch-and-bound method for finding exact solutions
Simulated annealing Genetic algorithms

Module code: 7CCSMPNN
Credit level: 7
Credit value: 15
Assessment:  written examination/s;  coursework; 

The aim of this course is to introduce both statistical and neural network theory and approaches for solving pattern recognition problems. Further, to consolidate lectures with MATLAB-based computer assignments.

At the end of the course students should:

• be familiar with Bayesian decision theory;
• be familiar with parametric density estimation;
• be familiar with nonparametric density estimation;
• be familiar with linear discriminant functions;
• be familiar with perceptrons;
• be familiar with Winner-take-all groups;
• be familiar with multi-layer perceptrons;
• be familiar with feature selection and extraction techniques;
• be familiar with clustering techniques.

Syllabus

• Introduction to pattern recognition
• Bayesian decision theory
• Parametric density estimation
• Nonparametric density estimation
• Linear discriminant functions
• Perceptrons
• Winner-take-all groups
• Multi-layer perceptrons
• Feature selection and extraction techniques
• Clustering technique