TU Delft
Year
print this page print this page     
NEDERLANDSENGLISH
Organization
2014/2015 Electrical Engineering, Mathematics and Computer Science Master Computer Science
IN4015
Neural Networks
ECTS: 6
Responsible Instructor
Name E-mail
Dr. J.A. Redi    J.A.Redi@tudelft.nl
Contact Hours / Week x/x/x/x
0/0/2/2
Pract.
Education Period
3
4
Start Education
3
Exam Period
4
5
Course Language
English
Expected prior knowledge
A basic knowledge of pattern recogntion and AI techniques is useful but not required. Basic programming skills are also useful.
Course Contents
In this course we will look at several techniques from the fields of Artificial Intelligence and Machine Learning that can help us to create adaptive systems that learn to act in complex, dynamic environments through interaction with that environment.

The emphasis will be on techniques that take their inspiration from biology.
Topics include: neural networks, evolutionary computation, learning classifier systems, swarm intelligence, artificial immune systems, subsumption architecture and embodiment.

The course is set up as a seminar, i.e. during the lectures we will discuss papers. As usual the lectures are split up in two ‘hours’ of 45 minutes. In each hour we will discuss one (sub)topic. The topic will first be presented by 1 or 2 people from the group (about 20 min.) after which we will discuss the topic with the whole group.

The second part of the course is a practical assignment. In groups of max. 3 people you will experiment with one or more of the techniques discussed during the lectures by implementing them in a (simulated) robot, e.g. to make a (group of) robots learn to navigate, to recognize or move certain objects or to communicate with each other. You can apply the technique you discussed during the lectures, but you may also use other techniques.
Study Goals
Upon successful completion of the course, students will be able to:

[LO1]. Describe the different bio-inspired techniques reviewed in the course, such as Neural Networks, Genetic Algorithms, or Learning classifiers.
[LO2]. Research literature concerning one of the above techniques, summarize it and report it to their peers (e.g., by means of a power-point presentation or a demonstration)
[LO3]. Debate upon positive and negative aspects of the techniques mentioned above
[LO4]. Implement one or more of the above mentioned techniques in a computer language (e.g. Java, C, C++, Html, Matlab scripts…)
[LO5]. Determine which technique(s) is most appropriate for being used in a certain problem domain, for example learning algorithms for robot navigation in unknown environments
[LO6]. Apply the appropriate technique to a (simple) problem domain (e.g., simulation of robot navigation in a simple maze)
Education Method
Seminar (including some lectures) and lab work.
Literature and Study Materials
Research papers that will be available through Blackboard.
Assessment
1. Presentation: during the lectures everyone presents one of the topics in the course.
2. Group assignment: you experiment with one or more of the discussed techniques in a (simulated) robotics platform and write a research report on you findings.