Syllabus

Credits: 5

Schedule: 09.01.2019 - 25.03.2019

Teacher in charge (valid 01.08.2018-31.07.2020): 

Prof. Rohit Babbar (academic year 2018-2019), Prof. Juho Rousu (academic year 2019-2020).

Contact information for the course (applies in this implementation): 

Course personnel

• Lecturer: Prof. Rohit Babbar
• Course assistants: Dr Sandor Szedmak, Viivi Uurtio, Eric Bach

Teaching Period (valid 01.08.2018-31.07.2020): 

III-IV (Spring)

Learning Outcomes (valid 01.08.2018-31.07.2020): 

After attending the course, the student knows how kernel methods can be used in various machine learning tasks, including classification, ranking and preference learning, as well as learning with multiple data sources and targets. The student knows how convex optimization methods can be used to efficiently train kernel-based models. The student knows how structured data such as sequences, hierarchies and graphs can be tackled through kernel methods.

Content (valid 01.08.2018-31.07.2020): 

Margin-based models and kernels. Classification and Support vector machines. Ranking and preference learning. Unsupervised learning with kernels. Kernels for structured data. Multilabel classification. Semi-supervised learning. Predicting structured output. Convex optimization methods.

Details on the course content (applies in this implementation): 

Inner product spaces, Kernels, Reproducing kernels, and RKHS. Introductory learning theory and Generalization. Optimization and Duality. Margin-based methods, Support vector machines, and SMO algorithm. Large-scale linear Optimization. PCA and  CCA. Kernels for structured data and structured prediction. Kernel mean embedding, characteristic kernels, and Maximum Mean Discrepancy. Random features

Assessment Methods and Criteria (valid 01.08.2018-31.07.2020): 

Exercises and exam.

Elaboration of the evaluation criteria and methods, and acquainting students with the evaluation (applies in this implementation): 

The course can be completed by two alternative ways:

• Exercises (max. 40 points + Bonus points) + Exam  (max. 40 points), giving a grade 0..5. Lowest passing points total is 40. 70 points will give the grade of 5.
• Exam only (max. 40 points), giving a grade 0...5. 20 points will give the grade 1, 34 points will give the grade of 5.

The better of the resulting two grades will be taken into account.

Workload (valid 01.08.2018-31.07.2020): 

Lectures and exercises

Study Material (valid 01.08.2018-31.07.2020): 

Shawe-Taylor and Cristianini: Kernel Methods for Pattern Analysis, Cambridge University Press, 2004. Slides and research papers provided during the course.

Substitutes for Courses (valid 01.08.2018-31.07.2020): 

ICS-E4030 Kernel Methods in Machine Learning

Prerequisites (valid 01.08.2018-31.07.2020): 

Bachelor's degree in computer science and the course CS-E3210 / T-61.3050 Machine Learning: Basic principles (or equivalent knowledge).

Grading Scale (valid 01.08.2018-31.07.2020): 

0-5

Description

Registration and further information