Machine Learning

• Explain the major steps involved in data modeling from preparing the data, modeling the data to evaluating and disseminating the results.
• Discuss key machine learning concepts such as feature extraction, cross-validation, generalization and over-fitting, prediction, curse of dimensionality, and the bias-variance trade-off.
• Match practical problems to standard data modeling problems such as dimensionality reduction, regression, classification, density estimation and clustering.
• Sketch how a relevant set of machine learning methods work and describe their assumptions, strengths, and limitations.
• Apply and modify machine learning algorithms in Python.
• Apply visualization techniques and statistics to evaluate model performance, identify patterns and data issues.
• Select, combine and modify data modeling tools in order to analyze data and disseminate the results of the analysis.
• Apply the data modeling framework to a broad range of application domains in medical engineering, bio-informatics, chemistry, electrical engineering and computer science.

Structured data modelling.
Data preprocessing and feature extraction.
Summary statistics.
Similarity measures.
Cost functions including maximum likelihood.
Optimization methods for machine learning.
Overfitting, generalization, regularization, and bias-variance tradeoffs.
Cross-validation.
Statistical evaluation and comparison of machine learning methods
Visualization and interpretation of models.
Dimensionality reduction (including principal component analysis).
Classification methods (decision trees, logistic and multinomial regression, nearest neighbor, naïve Bayes classifier, feed forward neural networks, and ensemble methods)
Regression methods (regression trees, linear regression, nearest neighbor, feed forward neural networks, and ensemble methods)
Clustering (k-means, hierarchical clustering, and mixture models)
Density estimation (kernel density estimation, Gaussian mixture models, and the EM algorithm)
Anomaly/outlier detection including density-based methods.

• 36 – 49 (Tues 13-17)

01001/01002/01003/01004/01005/02402/02403/02002/02101/02102/02525/02631/02632/02633/02692, Basic course in linear algebra and calculus, basic knowledge of probability theory and statistics, basic knowledge of Python.

The activities alternate between lectures, problem classes and hands-on Python exercises.

The course is a basic machine learning course relevant for all master programs. The course is designed as a stand-alone course and provides an introduction to basic machine learning, the mathematics behind the methods, and hands-on experience in their use.