Master
2024/2025
Research Seminar "Artificial Intelligence in Business"
Type:
Compulsory course (Master of Data Science)
Delivered by:
Big Data and Information Retrieval School
When:
2 year, 2, 3 module
Open to:
students of one campus
Language:
English
Course Syllabus
Abstract
The seminar is aimed at developing master's students' competences in organising and conducting research in the field of Data Science using modern methodological approaches. Within the framework of the course students will master the CRISP-ML(Q) methodology for structured construction of ML-projects, as well as MLOps principles for effective development and implementation of machine learning in industrial operation.
Special attention is given to the preparation of the final qualification work (FQW): from formulating a research question to presenting the results. Students will learn how to structure a research paper, work with sources, format the paper according to academic standards, create convincing visual presentations and successfully present the results of the research at a public defence.
As a result of the seminar, master's students will be able to independently conduct research in the field of Data Science and effectively present their findings.
Learning Objectives
- To develop a comprehensive understanding of the process of creating ML projects using the CRISP ML(Q) and MLOps methodologies for graduate students
- To prepare master's students for successful writing and defence of the final qualification work
- To teach methods of research planning and structuring of scientific work.
- To develop skills of searching and analysing the necessary literature on the topic of the research work.
- To develop skills of competent documentation and presentation of research results
- To teach to apply CRISP ML(Q) and MLOps methodologies in business projects.
- To prepare for public defence of the results of the research work
Expected Learning Outcomes
- Overview of CRISP-ML(Q) steps
- Specific applications for business problems
- Business Understanding and Data Understanding
- Modeling and Evaluation with Quality in mind
- Workshop: application of the methodology to graduate student topics
- Structure and requirements for the FQW
- Planning of work on the FQW
- Practicum: analysis of examples of successful projects
- Structure of the presentation for the defence
- Visualisation of research results
- Principles of effective data presentation
- Mistakes in presentations
- MLOps fundamentals
- Tools for reproducibility of experiments
- Versioning of data and models
- Automation of learning and evaluation processes
- Workshop: setting up a basic MLOps Pipeline
- Rules of the defence
- Preparation of the defence text
- Answering the questions of the committee
- Work with supervisor's feedback and review
- Workshop: mini defences with feedback
Course Contents
- CRISP-ML(Q) methodology in business projects
- Introduction to the final qualification work
- Preparing a presentation for the defence
- MLOps methodology
- Procedure of defence of the final qualification work
- Training defences
Interim Assessment
- 2024/2025 3rd moduleFinal Grade = Stage 1 (Chapter 1 of the FQW (Overview)+ beginning of Chapter 2 of the FQW (Problem Statement)) * 0.5 + Stage 2 (Chapter 1 FQW + Chapter 2 FQW +part of Chapter 3 (Software Implementation))* 0.5
Bibliography
Recommended Core Bibliography
- Машинное обучение. Паттерны проектирования: Пер. с англ. / В. Лакшманан, С. Робинсон, М. Мунн. - 978-5-9775-6797-8 - Лакшманан В. - 2022 - Санкт-Петербург: БХВ-Петербург - https://ibooks.ru/bookshelf/385740 - 385740 - iBOOKS
Recommended Additional Bibliography
- Машинное обучение. Портфолио реальных проектов. . - 978-5-4461-1978-3 - Григорьев Алексей - 2023 - Санкт-Петербург: Питер - https://ibooks.ru/bookshelf/390208 - 390208 - iBOOKS