F20DL - Data Mining and Machine Learning

Course leader(s):

• Yannis Konstas (Edinburgh, September)
• Dongdong Chen (Edinburgh, September)
• Neamat Elgayar (Dubai, September)
• Radu-Casian Mihailescu (Dubai, September)
• Abdullah Almasri (Malaysia, September)

Syllabus

1. Data sets (1.1 - Knowledge representation , 1.2 - Data Exploration and Visualization, 1.3 - Data preparation and feature selection)

2. Machine Learning workflow (2.1 - ML building blocks , 2.2 - Model Testing and evaluation , 2.3 - Hyperparameter tuning, 2.4 - Basic classifiers)

3. Unsupervised learning (3.1 - K means clustering, 3.2 - Cluster evaluation , 3.3 - Cluster visualization)

4. Supervised Learning (4.1 - Generative vrs discrinminative models, 4.2 - Bayes Classifier bayes rule , bayes inference, 4.3 - Decision Trees building decision trees, splitting criteria)

5. Linear Classification (5.1 - Linear Regression, 5.2 - Logistic Regression, 5.3 - Gradient descent rule)

6. Neural networks (6.1 - Single layer Perceptron , 6.2 - Multi-layer perceptron, 6.3 - Convolutional Neural networks)

Learning outcomes

By the end of the course, students should be able to do the following:

• construct key building blocks for data mining and machine learning including data preparation, feature extraction, model training and evaluation.
• differentiate between some popular machine learning approaches related to supervised and unsupervised learning
• practice the mathematics underpinning methodologies of data mining and machine learning.
• assess the benefits and limitations of using a variety of machine learning techniques for solving practical applications.
• choose appropriate solutions for practical problems based on complex issues like data quality, model complexity and overfitting.
• select a variety of software libraries to process data and evaluate machine learning pipelines.

Further details

Curriculum explorer: Click here

SCQF Level: 10

Credits: 15