Course detail

Artificial Intelligence

FEKT-MPC-UINAcad. year: 2026/2027

This course provides a systematic introduction to the field of Artificial Intelligence, spanning from classical state-space search methods to metaheuristics, evolutionary algorithms, and modern neural networks. Students will gain a solid understanding of fundamental AI principles and advanced optimization techniques, which form the backbone of intelligent decision-making.

The second half of the course focuses on Deep Learning and Artificial Neural Networks (ANN), the dominant force in AI development today. The curriculum covers everything from basic models to Transformer architectures, Large Language Models (LLMs), and diffusion models. The course provides a deep dive into the principles of Generative AI, autonomous agent learning mechanisms (RL), and the ethical boundaries of these technologies.

Graduates will acquire a broad and up-to-date overview of AI tools and trends, enabling them to navigate and apply AI effectively in a rapidly evolving technological landscape.

Language of instruction

Czech

Number of ECTS credits

5

Mode of study

Not applicable.

Guarantor

doc. Ing. Václav Jirsík, CSc.

Department

Department of Control and Instrumentation (UAMT)

Entry knowledge

Basic knowledge of linear algebra (matrix and vector operations), differential calculus, and multivariate analysis (gradients, derivatives), statistics, and fundamental algorithms and data structures is required. Basic proficiency in programming in any language is expected (Matlab, C). Prior knowledge of Python is not required but is considered an advantage.

Students are expected to maintain their own written notes reflecting the lecture material. Proficiency in this method of learning is considered a key component of the process of understanding and mastering the subject matter. In other words, the student is expected to have the skill to synthesize spoken word and visual input into their own coherent thoughts.

Rules for evaluation and completion of the course

To receive credit for this course, all of the following requirements must be met:

  • 100% attendance in the mandatory part of the tuition. Computer labs are compulsory; properly excused absences from computer labs may be compensated for by prior arrangement with the instructor.

  • Completion of two online training courses, verified by certificates.

  • Passing a progress test with a minimum of 5 points (max. 10 points).

  • Completion of 2 projects, each with a minimum of 5 points and a maximum of 10 points. Projects must be completed and defended within the specified timeframe.

  • Earned points will be added to the final exam grade.

The final exam is written, with a maximum of 70 points. A score below 35 points results in failing the exam. The final grade is determined by the sum of the exam points and the points earned during the practical sessions (labs/projects).

Aims

The aim of the course is to introduce students to selected methods and approaches in artificial intelligence (AI), covering both classical symbolic AI (GOFAI) and modern subsymbolic AI, with particular emphasis on neural networks. The objective of the course is thus to clarify the theoretical foundations of the discussed AI methods, to understand their implementation and application possibilities, and to develop the ability to critically evaluate the suitability of their use in practice.

Study aids

The course is primarily modeled after the UC Berkeley AI course and is based on the following publication: RUSSELL, Stuart and NORVIG, Peter. Artificial Intelligence: A Modern Approach. New Jersey: Pearson, 2021. 1170 pp. ISBN-13: 978-1-292-40113-3.

Study materials are provided selectively via the Moodle e-learning platform. Students are expected to attend lectures and maintain their own notes based on the material presented.

Prerequisites and corequisites

Not applicable.

Basic literature

MAŘÍK, Vladimír, ŠTĚPÁNKOVÁ, Olga, LAŽANSKÝ, Jiří a kolektiv. Umělá inteligence (1. až 6. díl) Praha: Academia 1993 - 2013. (CS)
RUSESELL, Stuart a NORVIG, Peter. Artificial Intelligence. A Modern Aproach. New Jersey: Prentice Hall 2010. 1132 s. ISBN-13: 978-0-13-604259-4. (EN)

Recommended reading

DUDA, Richard, HART Peter a STORK David. Pattern Classification. New York: John Wiley & Sons, INC. 2001. 654 s. ISBN 0-471-05669-3. (CS)
SONKA, Milan, HLAVAC, Vaclav a BOYLE, Rogert. Image Processing, Analysis and Machine Vision. Toronto: Thomson, 2008. 829 s. ISBN 978-0-495-24438-7. (CS)

Classification of course in study plans

  • Programme MPC-TIT Master's 0 year of study, winter semester, elective
  • Programme MPC-IBE Master's 2 year of study, winter semester, compulsory-optional
  • Programme MPC-EEN Master's 0 year of study, winter semester, elective
  • Programme MPC-EAK Master's 0 year of study, winter semester, elective

  • Programme MPC-AUD Master's

    specialization AUDM-TECH , 1 year of study, winter semester, compulsory-optional
    specialization AUDM-ZVUK , 1 year of study, winter semester, compulsory-optional

Type of course unit

 

Lecture

26 hours, optionally

Teacher / Lecturer

doc. Ing. Václav Jirsík, CSc.

Syllabus

  1. AI – Threat or Hope: Taxonomy, ethics, and AI safety.

  2. State Space I: Uninformed search strategies.

  3. State Space II: Informed search and heuristics.

  4. Adversarial Search or Expert Systems: Game playing (Minimax, Alpha-beta) or Expert Systems (Guest lecture: Assoc. Prof. Jirsík).

  5. Bio-Inspired Algorithms & Intro to Reinforcement Learning (RL): From evolution to agent-based learning (Agent-Env feedback loop).

  6. Advanced Metaheuristics: Surrogate models and optimization strategies.

  7. Foundations of ANN: Perceptron, ADALINE, and the principles of gradient descent.

  8. Specialized Architectures: Unsupervised learning, SOM (Kohonen maps), and RBF networks.

  9. Deep Learning: Backpropagation, MLP, regularization, and autoencoders.

  10. Convolutional Neural Networks (CNN): Architectures for computer vision and multi-class classification.

  11. Sequential Models and Attention: From RNN to Transformers – Why RNNs failed.

  12. Generative Models: LLMs, RLHF (Alignment), and Diffusion models.

  13. Course Colloquium: Buffer or selected advanced topics (SNN, GNN, PINNs, or Bayesian Networks).

Exercise in computer lab

26 hours, compulsory

Teacher / Lecturer

doc. Ing. Václav Jirsík, CSc.

Syllabus

The seminars are aligned with the lecture topics. The implementation will be carried out in Python and Matlab. Prior knowledge of Python is not required.