Machine Learning and Multivariate Statistics

Full course description

Machine Learning (ML) is the scientific study of algorithms and (multivariate) statistical methods that computer systems use to perform a task relying on patterns in the available data and inference. In its more general formulation, machine learning algorithms build a mathematical model in order to make predictions or decisions on new data. ML is related to computational statistics and uses methods and theories drawn from of mathematical optimization. Generally, ML tasks are classified into several broad categories. In supervised learning, a mathematical model is built from a set of data that contains both the inputs and the desired outputs (“training data”). Classification algorithms and regression algorithms are types of supervised learning. Classification algorithms are used when the outputs are restricted to a limited set of values. Regression algorithms are used when the output may have any value within a range. In unsupervised learning, a mathematical model is built from a set of data which contains only inputs and no desired output labels (also named Data Mining). Unsupervised learning algorithms are used to find structure in the data, like grouping or clustering of data points. Unsupervised learning can discover patterns in the data, and can group the inputs into categories, as in feature learning. Dimensionality reduction is the process of reducing the number of "features", or inputs, in a set of data.

          In System Biology, ML algorithms can aid in several ways. In data-rich scenarios (e.g. omics data) ML can help identifying the relevant pieces of information given a question, and how to make sense out of it (a "data mining" issue). Additionally, ML algorithms allow understanding how biological systems behave as the result of the integration and interaction between many individual components that can be monitored simultaneously. At the same time bioinformatics and systems biology have already induced significant new developments of general interest in ML, for example in the context of learning with structured data, graph inference, semi-supervised learning, system identification, and novel combinations of optimization and learning algorithms.

          In this course, ML will be applied in the context of neuroscientific investigations that on the basis of brain signals try to make sense of how the human brain processes information (visual auditory tactile) as well as in the analysis of omics data. You will study basic theoretical concepts of multivariate statistics and optimization that are fundamental to all ML algorithms. The problems of supervised (classification and regression) and unsupervised (clustering and dimensionality reduction methods) will be introduced formally together with common algorithmic solutions (e.g. Naïve Bayes, Support Vector Machines, Ridge Regression, [deep] neural networks, K-Means clustering). Practical examples of applications will be discussed highlighting best practices in the use of ML as well as in the communication of the obtained results. Finally, you will be tasked to use ML on available data to answer a specific question and report your results in the form of code and a written report.