Artificial Intelligence and Machine Learning

• 26 hrs lectures
• 13 hrs seminar
• 13 hrs projects

• 60 pts final exam (written part)
• 20 pts mid-term test (written part)
• 20 pts projects

Make students acquainted with the basics of artificial intelligence (AI) and machine learning (ML) that are the basic components of modern scientific methods, industrial systems and end-user applications - for example self-driving cars, cognitive robotics, recommendation systems, recognition of objects in images, chat-bots and many others. Show traditional techniques linked to currently dominating deep neural networks. Introduce basic mathematical formalism of AI and ML, that can be developed in specialized courses. Give an overview of software tools for AI and ML.

Students will:

• get familliar with basic nomenclature of machine learning, esp. of modern neural networks
• understand the relation between a task, a model and the process of learning
• review classical search-based methods of artificial intelligence and will see the possibilities of combining them with machine learning
• get familliar with basic machine learning models (gaussian models, gaussian classifiers, linear regression, logistic regression)
• get familiar with modern neural networks for solving different tasks (classification, regression, tasks in reinforcement learning scenarios) on various kinds of data (unstructured, image, text, audio) and with methods of their training

• Materiály k přednáškám dostupné v Moodlu
• C. Bishop: Pattern Recognition and Machine Learning, Springer, 2006
• Russel,S., Norvig,P.: Artificial Intelligence, Prentice-Hall, Inc., third edition 2010, ISBN 0-13-604259-7
• Ertel, W.: Introduction to Artificial Intelligence, Springer, second edition 2017, ISSN 1863-7310
• Goodfellow, I., Bengio, Y., Courville, A.: Deep Learning. MIT Press, 2016.

1. Introduction to artificial intelligence, machine learning and their relation
2. Basic tasks of machine learning (ML) - detection, classification, regression, prediction, sequence recognition, metrics for quality assessment.  
3. Probabilistic approach to classification and recognition - basics of Bayes theory. 
4. Gaussian model, its interpretation and training, PCA. 
5. Linear and logistic regression, Support vector machines - basic formulation and kernel trick.  
6. Neural networks (NN) - basic building blocks, principles of training.
7. Practical work with deep NNs - mini-batch, normalization, regularization, randomization, data augmentation.  
8. Image processing variants of NN: convolutional architectures, information aggregation, very deep networks
9. Sequentional variants of NN: RNN, LSTM, BLSTM, autoencoders, attention models, use of NN embeddings.
10. Reinforced learning with NNs and without them
11. State space search, game playing
12. Local search, constraint satisfaction problems
13. AI applications. 

Lectures will be immediately followed by demonstration exercises (1h weekly) where examples on data and real code in Python will be presented. Code and data of all demonstrations will be made available to the students.

The subject includes three homework assignments:

1. Data modelling and simple classifiers
2. Construction of a simple neural network
3. Problem solving by search

• Half-semestral exam (20pts)  
• Three homework assignments (20pts) 
• Semestral exam, 60pts, requirement of min. 20pts.