|
Lecturer(s)
|
-
Černý Antonín, Ing.
-
Zmeškal Ladislav, Ing. Ph.D.
|
|
Course content
|
The below itemized topics represent radii of the subject matter went through, they do not exactly correspond to scheduled lectures: 1. Introductory information, organization of the subject, recommended literature andsources of study materials; basic notion and definitions of the cognitive systems theory, relationship among data, information and knowledge, components of cognitive systems, general classification task. 2. Introduction into machine learning, supervised and unsupervised learning, applications and examples, case studies. 3. Bayes learning, Bayes theorem, optimal and naive bayesian classifier, hypothesis selection strategies, applications of NBC. 4. Linear regression, cost function derivation and techniques to minimize it, gradient descent derivation, gradient descent algorithm. 5. Multivariate linear regression, gradient descent in multidimensional space, problems and limitations of gradient descent; polynomial regression; normal equation. 6. Logistic regression, logistic regression hypothesis model, interpretation of the results, decision boundary, multi-class classification - One-vs-All algorithm. 7.Regularization, overtraining and its symptoms, techniques to avoid/suppress overtraining, naive derivation of regularization, regularization algorithm, regularized linear and logistic regression. 8. Support Vector Machines, optimization goal as an alternative perspective of logistic regression, mathematical model of SVM, hypothesis with safety factor, kernels. 9. Neural networks, history, biological pre-model of artificial neural networks, mathematical model of a neuron, MLP-type layered networks, classification via ANN, cost function of an ANN and its optimization, learning, Backpropagation algorithm. 10. Clustering, general remarks on unsupervised learning, K-means method, optimization criterion of the K-means, centroid selection, cluster number selection, K-means algorithm. 11. Dimensionality reduction, Principal Component Analysis, PCA functionality description and algorithm, PCA features, mathematical background of PCA, applications and case studies. 12. Blind source separation, motivation and definition of the blind source separation problem, Independent Component Analysis, ICA functionality description and algorithm, ICA features, mathematical background of ICA, applications and case studies.
|
|
Learning activities and teaching methods
|
Lecture supplemented with a discussion, Lecture with practical applications, Discussion, Laboratory work, Task-based study method, Individual study, Self-study of literature, Lecture with visual aids
- Graduate study programme term essay (40-50)
- 40 hours per semester
- Preparation for an examination (30-60)
- 30 hours per semester
- Presentation preparation (report) (1-10)
- 6 hours per semester
- Practical training (number of hours)
- 26 hours per semester
- Contact hours
- 39 hours per semester
- Preparation for laboratory testing; outcome analysis (1-8)
- 15 hours per semester
|
| prerequisite |
|---|
| Knowledge |
|---|
| prakticky využívat nabyté znalosti z oblasti umělé inteligence a rozpoznávání |
| prakticky využívat nabyté znalosti z matematické analýzy, lineární algebry, pravděpodobnosti a statistiky |
| kreativně aplikovat matematické poznatky s cílem nazírat na úlohy strojového učení jako na problém prohledávání N-dimenzionálního stavového prostoru |
| popsat pojmy a struktury teoretické informatiky, orientují se v základech výrokové i predikátové logiky |
| Skills |
|---|
| studovat odborné texty v anglickém jazyce |
| programovat na pokročilé úrovni v některém z vyšších programovacích jazyků, např. C++, C#, Object Pascal, SCALA, Java; praktická znalost MATLABu či Octave výhodou |
| Competences |
|---|
| N/A |
| learning outcomes |
|---|
| Knowledge |
|---|
| dosáhnout hlubšího pochopení základních technik strojového učení, reprezentace, odvozování a ukládání znalostí a racionálního chování, tj. rozhodování a řešení problémů |
| orientovat se v paradigmatech učících se systémů, zejména s přihlédnutím k jejich praktické aplikaci v oblasti umělé inteligence a inteligentního software |
| Skills |
|---|
| analyzovat existujiící algoritmy strojového učení a jejich teoretické specifikace v odborné literatuře |
| orientovat se v existujících implementacích učících se algoritmů, zejména s ohledem na jejich modifikaci, příp. optimalizaci |
| implementovat učící se algoritmy |
| zapojit se do řešení vědecko-výzkumných úkolů v oblasti umělé inteligence a strojového učení v rámci dalšího studia |
| Competences |
|---|
| N/A |
| teaching methods |
|---|
| Knowledge |
|---|
| Individual study |
| Self-study of literature |
| Task-based study method |
| Interactive lecture |
| Lecture supplemented with a discussion |
| Lecture with visual aids |
| Discussion |
| Skills |
|---|
| Lecture with visual aids |
| Competences |
|---|
| Task-based study method |
| assessment methods |
|---|
| Knowledge |
|---|
| Combined exam |
| Test |
| Skills |
|---|
| Seminar work |
| Competences |
|---|
| Combined exam |
|
Recommended literature
|
-
Barber, David. Bayesian reasoning and machine learning. Cambridge : Cambridge University Press, 2012. ISBN 978-0-521-51814-7.
-
Bishop, C.M. Pattern Recognition and Machine Learning. Springer, 2006. ISBN 978-0387-31073-2.
-
Heylighen, Francis. Cognitive Systems - A Cybernetic Perspective on the New Science of the Mind. Lecture Notes.. ECCO: Evolution, Complexity and Cognition. Vrije Universiteit Brusse, 2010.
-
Nilsson, J. Nils. Introduction to Machine Learning. Stanford University Press. Stanford University, 2005.
-
Trevor Hastie, Robert Tibshirani, Jerome Friedman. The Elements of Statistical Learning. Springer. 2009. ISBN 978-0-387-84857-0.
|