Lecture: Probabilistic Graphical Models


  • Lecturer(s):
  • Start: 20.10.2015
  • Dates: Tues. and alternating Thurs., 8:30 (s.t.) - 10:00, LBH / HS III.a
  • Course number: MA-INF 4315
  • Curriculum: Master
  • Effort: 6 CP
  • Exams: EXAM 2: 15.03.2016, 14:00 - 16:00, LBH / I.80; non-programming calculators allowed



This course introduces probabilistic graphical models and their use in solving problems in computer vision and machine learning. Graphical models offer a probabilistic framework for modelling and making decisions in complex scenarios with limited and noisy data.  We will cover topics such as Markov and Bayesian networks, inference techniques and parameter learning.  The theory will be demonstrated in computer vision applications such as human pose estimation, object tracking, image denoising and semantic segmentation.  

No prior knowledge of statistics is required to follow the course.  Exercises will be both theory and programming (Python) based and be completed in groups of two.


Additional Documents

Assignment Sheets

Exercise 1: ProbRefresh BayesianNetworks 291015
Assignment sheet  (PDF document, 351 KB)
Exercise 2: MarkovNetworksImgDenoise 121115
Assignment sheet  (PDF document, 173 KB)
Exercise 3: VE MsgPassing 261115
Assignment sheet  (PDF document, 64 KB)
Exercise 4: Sampling 101215
Assignment sheet  (PDF document, 412 KB)
Exercise 5: GraphCuts 140116
Assignment sheet  (PDF document, 334 KB)
Exercise 6: ParamLearning 210116
Assignment sheet  (PDF document, 205 KB)