SubjectsSubjects(version: 845)
Course, academic year 2019/2020
   Login via CAS
Applications of Computational Intelligence Methods - NAIL109
Title in English: Aplikace metod výpočetní inteligence
Guaranteed by: Department of Theoretical Computer Science and Mathematical Logic (32-KTIML)
Faculty: Faculty of Mathematics and Physics
Actual: from 2015 to 2019
Semester: winter
E-Credits: 6
Hours per week, examination: winter s.:0/4 C [hours/week]
Capacity: unlimited
Min. number of students: unlimited
State of the course: taught
Language: Czech
Teaching methods: full-time
Guarantor: Mgr. Roman Neruda, CSc.
Mgr. Martin Pilát, Ph.D.
Classification: Informatics > Informatics, Software Applications, Computer Graphics and Geometry, Database Systems, Didactics of Informatics, Discrete Mathematics, External Subjects, General Subjects, Computer and Formal Linguistics, Optimalization, Programming, Software Engineering, Theoretical Computer Science
Annotation -
Last update: T_KTI (05.05.2015)
Introduction of modern computational intelligence methods (evolutionary algorithms, machine learning and related fields) and their application to solving of real problems. Basic knowledge of machine learning, neural networks and evolutionary algorithms is required.
Aim of the course -
Last update: T_KTI (05.05.2015)

Teach advanced methods combining evolutionary algorithms, neural networks, and other computational intelligence methods. Deepen the knowledge from the introductory courses on neural networks, machine learning, data mining and evolutionary algorithms. The seminar will be focused on working with real data and the cooperation of various methods while solving difficult problems from the areas of optimization, learning and modelling.

Course completion requirements -
Last update: Mgr. Martin Pilát, Ph.D. (13.10.2017)

In order to pass the course, the student must obtain the credit, which is given for a presentation of group project created as part of the seminar.

Literature -
Last update: T_KTI (05.05.2015)

[1] Trevor Hastie, Robert Tibshirani, Jerome Friedman. The elements of statistical learning. Vol. 2, no. 1., Springer, 2009.

[2] Alex A. Freitas. Data Mining and Knowledge Discovery with Evolutionary Algorithms, Springer, 2002.

[3] Yoshua Bengio, Ian J. Goodfellow, and Aaron Courville. Deep Learning. MIT Press, 2015 (in print), [online: http://www.iro.umontreal.ca/~bengioy/dlbook]

[4] Charu C. Aggarwal: Data Mining - The Textbook. Springer 2015, ISBN 978-3-319-14141-1

[5] Thomas Bäck, Christophe Foussette, and Peter Krause. Contemporary Evolution Strategies. Springer Science & Business Media, 2013.

[6] P. Brazdil, C. Giraud Carrier, C. Soares, R. Vilalta: Metalearning. Springer, 2009.

Syllabus -
Last update: T_KTI (05.05.2015)

The subject aims at deepening of the knowledge from the following areas with the focus on their application to real data from different competitions (e.g. Kaggle, conference competitions, …).

Metalearning - model selection, hyper-parameter tuning (grid search, evolutionary algorithms), ensembles (bagging, boosting, stacking, blending)

Combination of evolutionary algorithms and machine learning - surrogate modelling; hybrid models, relation between local and global search, memetic algorithms; evolution and metalearning

Advanced evolutionary computing - CMA-ES, constrained optimization

Kernel methods - support vector machines (classification, regression), kernel neural networks, Radial Basis Function Networks

Semi-supervised learning - self-learning, generative models, Semi-Supervised Support Vector Machines, graph-based methods

Advanced models of neural networks - Echo State Network, Long Short Term Memory Network, autoencoders, convolution networks, Boltzmann machines, deep networks

 
Charles University | Information system of Charles University | http://www.cuni.cz/UKEN-329.html