About

The course covers the main tasks and mathematical methods of natural language processing.

Syllabus

• Preprocessing, feature extraction and classification.
• Vector representations of words.
• The task of marking sequences (tagging). The Linear-CRF model.
• Recurrent neural network models: GRNN, LSTM.
• Machine translation. The Sequence-to-sequence approach. The mechanism of attention.
• Transformer architecture.
• The task of language modeling.
• Statistical and neural network language models.
• The task of generating a natural language.
• Contextual vector representations of words.
• Transfer learning to NLP.
• The BERT model and its modifications.
• Text classification.
• Topic modeling and its applications.

Labworks

The course includes four practical tasks and an exam. The deadline for completing each task is 2 weeks. You can get up to 10 points for each task. For each day of delay, a penalty of 1 point is assigned.

Grading

0.7 * hw points / 5 + 0.3 * exam points.

Prerequisites

Machine learning, deep learning, Python.