This course describes the use of neural network models in learning and optimization e.g., pattern recognition, routing, and prediction. This course is divided into two parts. The first part (3 weeks) provides an introduction to neural networks, focusing on the so-called Hopfield model, its statistical mechanics and optimization algorithms. The second part (4 weeks) provides a more detailed introduction to learning, describing models, algorithms, and applications.
Student portal page: http://www.student.chalmers.se/sp/course?course_id=14209
External course webpage: None that currently is alive
The aim of this course is to give an understanding of fundamental concepts used to describe complex systems. As examples, chaotic low-dimensional systems, self-organizing systems, and cellular automata are discussed.
The fact that systems composed of a large number of simple components can exhibit complex phenomena is exemplified in, for example, self-organising systems (in the form of chemical reaction-diffusion systems), the second law of thermodynamics (as a statistical result of large physical systems), neural networks, evolution of cooperation, cellular automata (as an example of an abstract computational class of systems), economic systems of interacting trading agents, urban growth and traffic systems.
There is no universal definition of a complex system, but there are several features that are usually brought up when a system is considered complex, typically involving order/disorder and correlations. An important scientific question is whether these and other characteristics of the systems can be quantified in a comprehensive way. This is one of the aims with this course -- to provide a set of tools that can be used to give a quantitative description of a complex system for a variety of different types of systems. All quantities derived can be interpreted in terms of information, referring to the information quantity introduced by Shannon and Weaver.
Information-theoretic concepts can be applied on the macro-level of a system, for example, in order to describe the spatial structure formed in a chemical self-organizing system. The connection between information theory and statistical mechanics makes it possible to relate such an analysis to the thermodynamic properties and limitations of the system. Other examples include complex phenomena in cellular automata, entropy in microscopic physical systems (spin systems), and chaotic behaviour in dynamical systems.
Student portal page: http://www.student.chalmers.se/sp/course?course_id=14967
Course page:
The solution to problem 5.8 in the PDF below is updated with a minor addition regarding the criterion for the parameter α.
Note that the exam is on Friday, March 22, in the afternoon. (Previously, the last year's date was posted.)
In the third week we will bring up the 100 Floors Egg Dropping Puzzle. Thinking about coding may guide you to a solution of this problem.
We have posted a preliminary list of problems to be solved in the problem sessions below.
The second week we will introduce information theory for symbol sequences, which will serve as a key theoretical basis for several applications throughout the course.
An interesting problem will be discussed on Wednesday: The Monks Spots Puzzle. Try to solve that. Consider also the situation in an information perspective (which is indeed very tricky).
The course starts with an introductory lecture on Monday 21 January at 15.15 in MC.
All necessary information about the course is available below, including link to TimeEdit that contains the schedule, the lecture plan, the lecture nores in pdf, etc.
You may think about the balance problem puzzle that will be discussed in the first week.
The course provides an understanding of fundamental concepts used to describe complex systems, in particular dynamical systems such as chaotic low-dimensional systems, self-organizing systems, and simple spatially extended systems such as cellular automata. Many of the concepts are based in information theory.
The lectures will follow the presentation in:
K. Lindgren, Information theory for complex systems — An information perspective on complexity in dynamical systems, physics, and chemistry. (Chalmers, 2014.)
If you want to learn more: T. M. Cover and J. A. Thomas, Elements of information theory (Wiley, 1991).
See also: David MacKay, Information theory, Inference, and Learning (2003).
Kristian Lindgren (lecturer, examiner). Email: kristian.lindgren [at] chalmers.se
Susanne Pettersson (examples classes, projects) Email: susannep [at] chalmers.se
Rasmus Einarsson (examples classes, projects) Email: rasmus.einarsson [at] chalmers.se
The schedule is in TimeEdit.
Further details are given the lecture plan (pdf).
Preliminary list of problems to be solved in the problem sessions
Solutions
Five optional homework problems are given below. Each one gives up to two (2) extra points for the exam. Late submissions will normally not be graded. Hand-written solutions are fine, but please take care to make them legible.
Hand in your solutions in one of these ways:
The deadlines are:
Downloads:
Optional project work can be done in groups of 1-3 students. The project work is awarded up to 10 extra points for the exam. Further instructions are given in this file:
Project ideas and instructions for projects (pdf)
The exam is given on March 22, afternoon. A sheet with relevant equations etc is attached to the exam problems.
The course is graded based on the exam score including extra points from homework (max 10 points) and projects (max 10 points). The exam gives up to 50p. Grade limits (Chalmers/ECTS): 25p for 3/E, 28p for 3/D, 34p for 4/C, 38p for 4/B, 42p for 5/A. To pass, a minimum of 20p on the written exam is required, regardless of additional points.
Old exams, some of them with solutions (zip file with pdfs, about 17 MB) [updated 6 March 2018]
Exam 2018-03-16 problems (pdf) and solutions (pdf)
The course gives an introduction to the theory and description of nonlinear dynamical systems. How is chaos measured and characterized? How can one control and predict chaotic systems?
By using very simple mathematical models, it is possible to obtain chaotic behavior. Through the understanding of such systems, the students will get a feeling for how real systems can behave, learn how to characterize them and in certain special cases how to predict their behavior.
Student portal page: http://www.student.chalmers.se/sp/course?course_id=14968
External course webpage: http://fy.chalmers.se/~ostlund/dynamicalsystems
The second course in this field aims at giving a basic understanding of computational biology and theoretical models in molecular biology. This will include models of the origin of life, molecular evolution, and molecular genetics.
As a consequence of new measurement techniques, our knowledge of structure and function of biological macromolecules has increased significantly in recent years. The amount of data is now so large that it has become necessary to use computational and statistical methods of analysis. The new empirical data now allow statistically significant testing of models for genetic evolution. This has led to a renewed interest in evolution models on the genetic and molecular level. New numerical algorithms and mathematical models have been developed describing population genetics. It is the aim of this course to introduce the mathematical models and computational methods used in the analysis and modelling of genetical data and their evolution.
Student portal page: http://www.student.chalmers.se/sp/course?course_id=14473
External course webpage: http://fy.chalmers.se/~frtbm/ComputationalBiologyB/index.html
© 2019 FavThemes