Course Overview
This course integrates data-driven machine learning (ML) with principles of digital signal processing (DSP) and random processes, primarily through the least squares (LS) method.
Course objectives:
By the end of this course, students will be able to:
- Evaluate signal characteristics including amplitude, frequency, and phase.
- Analyze signals using auto-correlation and partial auto-correlation functions.
- Utilize AR, MA, and ARMA models for signal prediction (regression).
- Apply ARMA models to multivariate exogenous and endogenous signals.
- Compare and select appropriate performance metrics.
- Classify signals using logistic regression and feature engineering techniques.
Special Notes:
- This is a newly developed course tailored to be self-contained, with no formal prerequisite courses required. However, a basic understanding of probability and linear algebra is assumed.
- The course emphasizes hands-on learning, with assignments conducted using Python.
Grading Policy
The course grade consists of homework assignments (100%).
Homework
- Coding assignments in Matlab.
- Submission via the Moodle page.
- The homeworks are graded based on the following criteria:
- Solution correctness (25%)
- Code quality (25%)
- Explanation clarity (25%)
- Formatting (25%): special attention is required for the following:
- Plots are required to be labeled and have a title
- Headings for each section
- Submissions should include a LiveScript notebook and corresponding PDF file.