Arcadia AI — Master's Program
EN
РусскийO'zbekchaEnglishDeutschSrpskiहिन्दी
Program  /  Semester 1 — AI Foundations
Course

Classic ML, Introduction to Deep Learning

From linear models to neural networks: classic ML and first steps into Deep Learning

NSDNYL
Nikita Severin, Dmitry Nesterov, Yaroslav Leonov
Faculty roster still under discussion

About the course

The course builds an end-to-end understanding of machine learning — from regression and SVMs to convolutional and recurrent networks. It pays special attention to generalization theory, the bias-variance tradeoff, and proper model validation. It establishes the shared vocabulary and experimental foundation for every subsequent course in the program.

What you'll learn

✓Build and validate classic ML models with well-reasoned hyperparameter choices
✓Implement neural networks from scratch, understanding the role of every training component
✓Diagnose underfitting and overfitting and apply the appropriate regularization techniques
✓Compare algorithms on their theoretical and empirical generalization properties

Key topics

•Linear and logistic regression; regularization (Ridge, Lasso, ElasticNet)
•Quality metrics: classification, regression, ranking
•Decision trees, ensemble methods (Random Forest, Gradient Boosting)
•SVMs and kernel methods
•Generalization theory: VC dimension, PAC learning
•Backpropagation and automatic differentiation
•Convolutional neural networks (CNNs): architectures, pooling, transfer learning
•Recurrent networks (RNN, LSTM, GRU)
•Normalization (Batch Norm, Layer Norm), Dropout, early stopping
This description was generated automatically and has not yet been reviewed by an instructor — it's a draft for discussion.