Artifical Inteligence (FSI-RAI)

Academic year 2020/2021
Supervisor: doc. Ing. Jiří Krejsa, Ph.D.  
Supervising institute: ÚMTMB all courses guaranted by this institute
Teaching language: Czech
Aims of the course unit:
Understanding of the basics of artificial intelligence approaches and ability to apply those in solving engineering tasks.
Learning outcomes and competences:
Student will gain the overal knowledge in the area of artificial intelligence methods and will be capable of applying the appropriate methods in solving engineering problems.
Prerequisites:
Vector and matrix calculations, algoritmization abilities, ability to implement given algorithm in Matlab and/or Python.
Course contents:
The course introduces the essential approaches in artificial intelligence area, including the state space search methods, stochastic optimization and machine learning, in particular the artificial neural networks including the convolution neural networks. Usage of the methods is demonstrated on solving simple engineering problems using corresponding tools (Matlab, TensorFlow).
Teaching methods and criteria:
The subject is taught using a set of lectures explaining the basic principles and theory of given area. The practical part is focused on actual implementation of explained methods using appropriate tools, such as Matlab or Python.
Assesment methods and criteria linked to learning outcomes:
The subject evaluation is based on the implementation of softttware project that uses selected method of artificial intelligence. The project final report has to be delivered including the source code and presented to the audience in the form of short presentation.
Controlled participation in lessons:
Lectures are optional, but are highly recommended. The practices are obligatory. The way the student can substitute its absence is up to the teacher.
Type of course unit:
    Lecture  13 × 2 hrs. optionally                  
    Computer-assisted exercise  13 × 2 hrs. compulsory                  
Course curriculum:
    Lecture 1. Introduction, areas of artificial intelligence.
2. State space search - introduction.
3. Blind and informed methods of state space search.
4. Game theory – min/max algorithm
5. Evolution methods of state space search.
6. Basic paradigms of neural networks
7. Unsupervised/supervised learning.
8. Backpropagation.
9. Approximation versus classification.
10. Convolution neural networks - intro
11. Convolution neural networks - topology, convolution and pooling layers
12. Reinforcement learning
13. Q-learning
    Computer-assisted exercise 1. Essential tools: Matlab, Python, Tensor Flow, Keras.
2. Breadth/depth first search algorithms
3. Dijkstra algorithm, A-star
4. Min-max algorithm
5. Genetic algorithm
6. Layered networks, Neural Network Toolbox
7. Layered networks – examples
8. Convolution neural network – Tensor Flow
9. Reinforcement learning and Q-learning
10. Project, consultations
11. Project, consultations
12. Project, consultations
13. Project presentation
Literature - fundamental:
1. Mařík a kol.: Umělá inteligence (1-6), Academia
Literature - recommended:
1. Hope T.: Learning TensorFlow: A Guide to Building Deep Learning Systems, O'Reilly Media, 2017
The study programmes with the given course:
Programme Study form Branch Spec. Final classification   Course-unit credits     Obligation     Level     Year     Semester  
N-MET-P full-time study --- no specialisation -- Cr,Ex 5 Compulsory 2 1 S