Examination arrangement

Course content

This course provides more comprehensive content on machine learning (ML) principles and techniques for developing practical ML systems. Topics covered include essential ML basics, modern prediction models, with a focus on ensemble learning, and critical steps in the ML pipeline, such as data preparation, manipulation, exploratory data analysis, data cleaning, feature engineering, and model interpretation. The course also addresses model evaluation, reproducibility, automatic ML, and specialized methods for time series data.

Learning outcome

By the end of the course, students will be prepared to develop robust machine-learning systems for real-world applications.

Learning methods and activities

Lectures, group work, colloquia, and self-study.

Further on evaluation

The course evaluation consists of two components:

1. Individual Assignment (IA): Approximately one month after the course commences, every student is required to complete an individual assignment (IA). Each student has the opportunity for a second attempt to pass the IA; however, in the event of a second attempt, a 5% deduction in course points will be applied. Students who do not pass the IA in both attempts will receive a final course grade of 'F' or 'Fail.'
2. Course Project: Only those students who successfully pass the IA are eligible to proceed to the course project. Course projects are graded as a team effort, with each team comprising a maximum of three students. Project grading is determined by combining base points (ranging from a maximum of 100% to a minimum of 41%) with potential project deductions (ranging from 0% to -17%). Base points are proportional to the number of Virtual Teams (VTs) that the student team outperforms in terms of prediction performance. VTs are prepared by the instructors and teaching assistants. If a student team fails to outperform any VT, all team members will fail the project, resulting in an overall course failure. Potential deductions may include late project submissions (within three days) and inadequate documentation of key machine learning practice components.

Final course points are rounded to letter grades based on NTNU standard ranges. In the event that a student receives a final grade of 'F' or 'Fail,' they will be required to retake the entire course.

Recommended previous knowledge

TDT4136 Introduction to Artificial Intelligence, TDT4171 Artificial Intelligence Methods, TDT4172 Introduction to Machine Learning or similar.

Course materials

Textbooks:

• Tom Mitchell: Machine learning, McGraw Hill, 1997.
• Christopher M. Bishop: Pattern Recognition and Machine Learning, 2006
• Dipanjan Sarkar, Raghav Bali, Tushar Sharma: Practical Machine Learning with Python: A Problem-Solver's Guide to Building Real-World Intelligent Systems, 2017

Selected papers and code examples.

Credit reductions