Deep Learning Systems

The aim of the course is to provide the student with theoretical and practical knowledge of artificial neural networks. The content includes classical machine learning algorithms and deep artificial neural networks for classification and prediction, feature extraction problems. Students will learn unsupervised, supervised, reinforcement learning algorithms in classical machine learning methods and artificial neural networks: multilayer perceptron, convolutional neural networks, recurrent neural networks, deep-Q networks. Training algorithms and methods to prevent overtraining will be analyzed. Students will learn to design and built classical machine learning and?artificial neural network systems to solve real world tasks.

The aim of this course is to provide knowledge about deep neural network methods and classical machine learning algorithms and develop the ability to apply these methods to solving real problems.

Basics in mathematics and programming

1. History and recent achievements in artificial intelligence development. Ethics and artificial intelligence. 2. Data preparation. 3. Types of learning: supervised, unsupervised, reinforcement learning. 4. Supervised learning: linear regression, binary logistic regression, decision tree. Assessment of classification and prediction accuracy. 5. Unsupervised learning: k-means clustering, hierarchical clustering, principal component analysis. 6. Methods of feature selection. 7. Neural networks and error back-propagation training algorithm. 8. Convolutional neural networks. 9. Recurrent neural networks for sequence analysis. 10. Reinforcement Learning: Deep Q-networks. 10. Neural network training algorithms. Methods to prevent overtraining. 11. Complex structures of artificial neural networks.

1. Explains classical machine learning algorithms and artificial neural network methods, data preparation, feature extraction algorithms. 2. Explains artificial neural training algorithms. 3. Designs and implements classic machine learning algorithms and artificial neural network systems, evaluates accuracy, interprets results.