An overview course on basic neural network models and their application to image, text, and sound processing tasks.
- Feed-forward networks. Matrix-vector differentiation. Automatic differentiation.
- Optimization methods for neural networks. Regularization and DropOut.
- Batch normalization, convolutional neural networks.
- Semantic segmentation. Object detection.
- Neural network image stylization.
- Recurrent neural networks.
- Reinforcement learning. Q-training. The DQN model.
- Algorithms of Reinforce, A2C.
- Generative-adversarial networks.
- Variational auto-encoder.
- Parameterization methods. Variational auto-encoder with discrete variables.
5 practical tasks and 1 lab work. Each task is evaluated from 10 points, and the lab - from 5 points. The exam is evaluated from 10 points.
0.7 * semester points + 0.3 * exam points.
Machine learning, statistics, optimization methods.