Introduction to neural network and machine translation

2023/2024

Учебный год

ENG

Обучение ведется на английском языке

8

Кредиты

Кто читает:

Кафедра прикладной математики и информатики (Нижний Новгород) (Факультет информатики, математики и компьютерных наук (Нижний Новгород))

Статус:

Курс по выбору

Когда читается:

4-й курс, 2, 3 модуль

Преподаватели

Орлов Михаил Максимович

Поздняков Виталий Витальевич

Course Syllabus

Abstract

The course introduces basic concepts of neural networks, deep learning and machine translation.

Learning Objectives

• The purpose of the ciyrse is to develop the ability to use neural network in their research and applied projects.

Expected Learning Outcomes

• Is able to use word embedding models
• Is able to use supervised learning
• Understands the advantages and disadvantages of neural networks
• Can create and use convolutional neural networks
• Can create and use recurrent neural networks
• Can create and use attention-based neural networks
• Can pretrain and fine-tune neural networks and their components
• Understands the principles of large language models and knows how to use them to solve applied problems

Course Contents

• Word embedding, word2vec model
• Supervised learning, logistic regression, multilayer perceptron
• Overfitting problem, regularization
• Convolutional neural networks
• Recurrent neural networks, Seq2seq modeling
• Attention-based models, Transformers
• Pretraining and fine-tuning, BERT, GPT
• Large language models, Prompt engineering, Chain-of-thought

Assessment Elements

• Lab 1
• Lab 2
• Lab 3
• Lab 4
• Project
• Quizzes
  Test tasks that are given at every lecture except the first.

Interim Assessment

• 2023/2024 3rd module
  Score = 0.15*Lab 1 + 0.15*Lab 2 + 0.2*Lab 3 + 0.3*Lab 4 + 0.1*Project + 0.1*Quizzes Final score = 0.8*Score + 0.2*Bonus Bonus: A bonus point is awarded if a student has voluntarily gone beyond the scope of the discipline. An example of such work is the application of a method considered as part of an individual project on other data, or the adaptation of a basic model from laboratory 4 with ideas from a recent research on neural networks. The work that can be evaluated for a bonus point is previously agreed with the teacher.A student gets a maximum score of 10 if he/she has successfully completed at least two works that deserve a bonus.

Bibliography

Recommended Core Bibliography

• Kelleher, J. D. (2019). Deep Learning. Cambridge: The MIT Press. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edsebk&AN=2234376

Recommended Additional Bibliography

• Ian Goodfellow and Yoshua Bengio and Aaron Courville. Deep Learning, 2016. URL: http://www.deeplearningbook.org
• Ian Goodfellow, Yoshua Bengio, & Aaron Courville. (2016). Deep Learning. The MIT Press.
• Николенко С., Кадурин А., Архангельская Е. - Глубокое обучение. — (Серия «Библиотека программиста») - 978-5-4461-1537-2 - Санкт-Петербург: Питер - 2020 - 377026 - https://ibooks.ru/bookshelf/377026/reading - iBOOKS