Multi-valued Logic Applications (FSI-SAL)

The aim of the course is to introduce students to the mathematical background of artificial intelligence methods and also to teach them how to implement these methods with understanding.

The areas that will be covered in the course, which students will study and program:

1. Nearest Neighbor Method, Decision Trees, Support Vector Machine.

2. Building a neural network for training on tabular data.

3. Convolutional neural networks for working with image data.

4. R-CNN for detecting a particular object in images.

5. Autoencoders and decoders.

Programming skills in Matlab, statistical methods.

The course is devoted to artificial intelligence algorithms in both theoretical and practical aspects. In the course, students will learn about the theoretical mathematical background of each area of methods and then implement them. Matlab is used as the programming environment and some implementations will be presented in Python.

The first part of the course covers machine learning methods - kNN, Support Vector Machine, decision trees. The second part discusses various neural networks, deep learning and more complex R-CNNs and autoencoders. Students will learn how to create their own training and test data, build appropriate layers such as convolutional neural networks, perform validation and evaluation of results.

The course also includes invited lectures on language analysis using neural networks.

1. Relationship of artificial intelligence methods to expert systems.

2.-3. Machine learning methods (kNN, decision trees, SVM, etc.).

4.-5. Basic neural network design for tabular data, explanation of back-propagation.

6. Convolutional neural networks (convolution, pooling, batch normalization).

7. Autoencoders and decoders.

8.-9. Invited talk on natural language processing, chatbots.

10. R-CNN (convolutional neural network for image retrieval), transformers.

11.-12. Work on semester project, tutorials.

13. Presentation of final projects, evaluation.

1. Design of expert system in Matlab (connection with fuzzy logic).

2.-3. Implementation of kNN, decision trees, SVM methods on different data. Test and validation data.

4.-5. Design of neural networks for prediction on given data (e.g. medical data, economic indicators, etc.).

6. Processing of image databases for designing convolutional neural networks (recognition of handwritten digits, geometric shapes, animals).

7. Autoencoders and decoders - implementation for noise reduction, image retrieval, data dimensionality reduction.

8.-9. Chatbot design, working with ChatGPT.

10. R-CNN on real data.

11.-12. Semester project consultation.

13. Presentation, evaluation of work.

KIM, Phil. MATLAB Deep Learning: With Machine Learning, Neural Networks and Artificial Intelligence. Berkeley, CA: Apress, 2017. ISBN 978-1-4842-2845-6.

Neural Networks and Deep Learning. Online. Michael Nielsen, 2015. Dostupné z: http://neuralnetworksanddeeplearning.com/. [cit. 2023-10-30].