It is assumed that the student is already familiar with the basic neural networks (NN), simpler NN architectures and principles for estimation and prediction of simpler NN models.
Main field of studyComputer Science
Course typeProgramme course
Course coordinatorAnders Eklund
Director of studies or equivalentJolanta Pielaszkiewicz
|Course offered for||Semester||Weeks||Language||Campus||VOF|
|F7MSL||Statistics and Machine Learning, Master´s Programme||2 (Spring 2020)||v202014-202023||English||Linköping||o|
Main field of studyComputer Science
Course levelSecond cycle
Course offered for
- Masters Programme in Statistics and Machine Learning
- Bachelor’s degree equivalent to a Swedish Kandidatexamen in one of the following subjects: statistics, mathematics, applied mathematics, computer science, engineering
- Completed courses in calculus, linear algebra, statistics, and programming
- A course in machine learning that covers at least 6 ECTS and includes neural networks
- English corresponding to the level of English in Swedish upper secondary education (English 6/B).
- Exemption from Swedish 3/B
Intended learning outcomes
After completion of the course the student should at an advanced level be able to:
- use relevant concepts and methods from Deep Learning in order to formulate, structure and solve practical problems that involve large and complex data
- choose a deep learning architecture that is appropriate for a given data structure, problem formulation and application area
- choose appropriate activation functions and hyperparameter settings in Deep Learning models
- estimate the performance of Deep Learning models
- use existing Deep Learning software in order to analyze large and complex datasets, tune the network architecture and make predictions.
The course introduces main concepts in Deep Learning and widely used Deep Learning models. The course includes the following topics:
- Deep and shallow networks
- Regularization, droupout and early stopping. Optimization of deep neural networks
- Convolutional neural networks and image analysis
- Deep recurrent neural networks and sequence analysis
- Autoencoders and feature extraction
- Generative Adversarial neural networks
Teaching and working methods
The teaching comprises lectures, practical sessions and computer exercises complemented by self-studies. lectures are devoted to presentations of theories, concepts and methods. Practical sessions are devoted to presentations of practical tools needed for computer exercises. Computer exercises provide practical experience of data analysis with Deep Learning software.
Written reports on the computer assignments.
Students failing an exam covering either the entire course or part of the course twice are entitled to have a new examiner appointed for the reexamination.
Students who have passed an examination may not retake it in order to improve their grades.
Planning and implementation of a course must take its starting point in the wording of the syllabus. The course evaluation included in each course must therefore take up the question how well the course agrees with the syllabus.
The course is carried out in such a way that both men´s and women´s experience and knowledge is made visible and developed.
DepartmentInstitutionen för datavetenskap
|LAB1||Laboratory work||EC||3 credits|
This tab contains public material from the course room in Lisam. The information published here is not legally binding, such material can be found under the other tabs on this page. There are no files available for this course.