FYS-3012 Pattern recognition

Emnebeskrivelse

Work Load 10  credits (ECTS)

Type of course Theoretical subject.

Level of course/Year of study Master level course.

Language of instruction Norwegian, or English by request

Examination semester Autumn, Spring

Contents The students are assumed to have fundamental knowledge of probability theory, signal processing and programming (MATLAB or IDL). The subject deals with supervised and unsupervised classification of feature vectors (measured data). Included in the course are Bayes classification (statistical), Linear classifiers (the Perceptron rule, minimization of MSE), nonlinear classifiers (multilayer Perceptron, RBF- network), clustering (sequential and hierarchical methods, K-means) and feature selection/generation.

Recommended previous qualifications FYS-2006 Signal processing, STA-1001 Probability and statistics 1

Form of teaching Lectures: 45 hr Exercises: 45 hr

Coursework requirements Exercises

Exam and assessment Portfolio assessment of a two weeks¿ written home examination (counting approx 25%) and a final oral examination (counting approx 75%). Grading scale: A-F. Re-sit examination: Early in tutor free term, with reserve of postponed examination oral examination which counts 100%. Postponed examination: According to the current regulation of exams § 21. New examination: According to the current regulation exams § 25, last paragraph.

Contact person/lecturer Arvid Aanstad, tlf: +47 77 64 40 06, e-mail: stud-kons@phys.uit.no

Recommended reading/syllabus Available on request

Course overlap FYS-363 -1      9 sp FYS-8012 -1      10 sp

Homepage http://uit.no/fysikk/fys3012/

Timetables http://timeplan.uit.no/emne.php?year=2006&department=ALLE&module=FYS-3012-1