Free study materials
Neural networks textbook
We are making the study texts from our Machine Learning and Neural Networks course freely available. We publish selected chapters – each as a separate PDF, from the basics of machine learning through convolutional and generative networks to natural language processing. The texts are in Czech.
Selected chapters to download
We publish selected chapters of the textbook (in Czech). Each is a separate PDF with chapter goals, key ideas and step-by-step explanations; we add further chapters over time.
-
Machine learning
Study materials – Introduction to machine learning (Czech)
What a model is and how it is created, supervised, unsupervised and reinforcement learning, the difference between classification, regression and clustering, training, validation and test data, metrics and the problem of overfitting.
-
Decision trees
Study materials – Decision trees (Czech)
The principle of decision trees, entropy and the Gini index, implementation in scikit-learn, visualising and interpreting a tree, hyperparameters and their tuning (GridSearchCV, RandomizedSearchCV), an introduction to Random Forest and Gradient Boosting.
-
Feedforward neural networks
Study materials – The perceptron and the basics of deep learning (Czech)
The perceptron as the basic building block of neural networks: weights, bias and activation functions, training on data, the limits of linear separability and the path to multi-layer networks.
-
Convolutional neural networks
Study materials – Convolutional neural networks (Czech)
Why classic networks are not enough for images, an image as a grid of numbers, convolutional filters and the main CNN layers, the journey of an image through the network to a decision, and areas of practical use.
-
Data augmentation and transfer learning
Study materials – Transfer learning (Czech)
Transferring knowledge from pre-trained models: training from scratch vs. transfer learning, fine-tuning and feature extraction, freezing layers and adjusting the classification head (PyTorch, ResNet).
-
Generative networks
Study materials – Generative artificial intelligence (Czech)
How generative models imitate data and create new samples: the difference between generative and discriminative models, modelling data distributions, step-by-step generation of text and images, and examples of use.
-
Natural language processing
Study materials – Natural language processing (NLP) (Czech)
From raw text to understanding: tokenisation and preprocessing, embeddings and vector representations of words, static vs. contextual representations, bias in training data and the attention mechanism (self-attention).
Terms of use
The textbook is published by Turing Institut (Bee Learning s.r.o.). Anyone may download and read it for their own study free of charge. Commercial use and use for any teaching other than that of Turing Institut is not permitted.
When citing, please credit the source turinginstitut.cz. Interested in another kind of use? Write to us.
How to work with the textbook
Read chapter by chapter
Each chapter is a separate PDF with goals, key ideas and step-by-step explanations. Start with the basics of machine learning, or pick just the topic you need right now.
Theory from our course
The texts were written as study support for the Machine Learning and Neural Networks course. In the lessons they are followed by practical exercises in Python, Jupyter notebooks and work with a lecturer.
Continuously updated
For now we publish selected chapters of the textbook. We gradually add more and update the texts as the field evolves – it is worth coming back from time to time.
Want to go deeper?
The textbook is the study support for the Machine Learning and Neural Networks course. In the course it is followed by live lessons with a lecturer, practical tasks in Python and feedback on your solutions.