Магистратура
2024/2025
Введение в глубинное обучение
Статус:
Курс обязательный (Магистр по наукам о данных)
Направление:
01.04.02. Прикладная математика и информатика
Где читается:
Факультет компьютерных наук
Когда читается:
2-й курс, 2 модуль
Формат изучения:
с онлайн-курсом
Онлайн-часы:
60
Охват аудитории:
для своего кампуса
Преподаватели:
Саночкин Юрий Ильич
Прогр. обучения:
Магистр по наукам о данных (о)
Язык:
английский
Кредиты:
3
Course Syllabus
Abstract
This course introduces the students to the elements of machine learning, including supervised and unsupervised methods such as linear and logistic regressions, decision trees, support vector machines, bootstrapping, random forests, boosting, regularized methods. Students will apply Python programming language and popular packages, such as pandas, scikit-learn, to investigate and visualize datasets and develop machine learning models that solve theoretical and data-driven problems. Pre-requisites: at least one semester of calculus on a real line, vector calculus, linear algebra, probability and statistics, computer programming in high level language such as Python.
Learning Objectives
- The course aims to help students develop an understanding of the process to learn from data, familiarize them with a wide variety of algorithmic and model based methods to extract information from data, teach to apply and evaluate suitable methods to various datasets by model selection and predictive performance evaluation.
Expected Learning Outcomes
- Form an understanding of the core tools of the course
- Gained experience in learning, presenting, reviewing and discussing a paper, a deep understanding of the NeRF / Word2Vec overview
- Gained experience in learning, presenting, reviewing and discussing a paper, a deep understanding of the Attention Is All you need / Model-agnostic meta-learning for fast adaptation of deep networks
- Gained experience in learning, presenting, reviewing and discussing a paper, a deep understanding of the CLIP / GAN
- Gained experience in learning, presenting, reviewing and discussing a paper, a deep understanding of the Typical Decoding / NeRF
- Gained experience in learning, presenting, reviewing and discussing papers, a deep understanding of the Robustness May be at Odds with Accuracy
- Construct machine learning models on the proposed data sets in Python.
- Evaluate performance of the models.
- Build features suitable for the selected machine learning models.
- Tune models to improve prediction and classification performance of the models.
Course Contents
- Introduction
- Papers: NeRF / Word2Vec
- Papers: Attention Is All you need / Model-agnostic meta-learning for fast adaptation of deep networks
- Papers: Typical decoding / GAN
- Papers: How Can We Know What Language Models Know? / NeRF
- Papers: Robustness May be at Odds with Accuracy
Assessment Elements
- Average test mark
- Average review markThe review mark is averaged along all weeks when the student writes a review. The penalties for submitting the review after the deadline are progressive: 0.2 is the base penalty for submitting late. Every 24 hours it increases by 0.1 until the review is submitted. So for the review deadline on Thursday at 23:59, submitting it on Friday at 00:01 would be fined 0.2 of the mark, on Saturday 0.3 of the mark, on Sunday 0.4 of the mark, and so on.
- Presentation markThe Presentation mark is the average of the marks assigned during peer review.
- Final Test
Interim Assessment
- 2024/2025 2nd module0.1 * Average review mark + 0.3 * Average test mark + 0.3 * Final Test + 0.3 * Presentation mark
Bibliography
Recommended Core Bibliography
- Eco, U., Farina, G., & Mongiat Farina, C. (2015). How to Write a Thesis. Cambridge, Massachusetts: The MIT Press. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edsebk&AN=963778
Recommended Additional Bibliography
- Mikael Sundström. (2020). How Not to Write a Thesis or Dissertation. Edward Elgar Publishing.