ML 2024 (MS)#

Course assessment#

Activity

Final scores

Attendance and participation

\(5\%\)

Practice/Quiz

\(20\%\)

Homework/SIS

\(20\%\)

Mid-term

\(15\%\)

Final project

\(40\%\)

Quizzes are held during classes, usually devoted to the topic of the previous lecture.

Homeworks assessments (SIS) consists of assignments in Jupyter Notebooks.

Mid-term: Kaggle competition and project proposal.

Final project: application of machine learning to a real-world problem.

Create a group project that solves a real-world problem based on machine learning algorithms. Recommended team size is 3-4 students. Other guides see on DL course page.

Project defense guidelines#

All teams are required to present their projects in front of the audience. Time for presentation is 10 minutes (+5 minutes for QA). Also, a report in Jupyter Notebook should be provided with all the code and results. Note that the report should be consice and consistent, equipped with sufficient number of formulas and graphs, without tons of meaningless general words and other irrelevant stuff.

Assessment criteria

  • Actuality and relevance: how important and relevant is the problem that you are solving?

  • Novelty and originality: how unique and impactful is the problem that you are solving?

  • Related work: how much related work you have found and how you are going to improve it?

  • Presentation quality: how good are your slides and how well you present your work?

  • Technical details: how did you implement the solution? What methods, algorithms, metrics, datasets did you use? Why did you choose them?

  • Results: how good is the performance of your model?

  • Visualization: how clear and insightful are your charts or dashboards?

  • Visibility: how accessible and easily reviewable is your work? Is there a testable prototype available on the Web?

  • Role of members: is the role of each teammate is clearly understandable?

  • Report and source code: how concise is your report? Is it written in a single .ipynb file? Does it contain all necessary explanations, formulas and code (the latter could be in the form of links to GitHub or alike)?