Dozent

Dr. Matthias Rupp

Fachbereich: Informatik und Informationswissenschaft, Universität Konstanz

Termine (Vorlesung)

Mo 13:30 - 15:00 R 711

Inhalt

The course consists of two blocks, a first block taught by Dr. Sutter and a second block taught by Dr. Rupp. The contents of the two blocks are

Block 1

  • Introduction (e.g., forms of learning, motivating examples, basic concepts)
  • Regression and classification (e.g., linear regression, overfitting, cross-validation/bootstrap, model selection, bias-variance tradeoff)
  • The statistical perspective (e.g., frequentist vs. Bayesian statistics, regularization as a prior, loss as likelihood)
  • Statistical decision theory (e.g., optimization problems, decision making based on statistical models and utility functions)
  • Unsupervised learning (e.g., k-means, principal component analysis, other forms of dimensionality reduction)

Block 2

  • Kernel-based learning (e.g., the kernel trick, properties of kernels, kernel ridge regression, kernel principal component analysis, support vector machines)
  • Neural networks (e.g., perceptron, backpropagation, stochastic gradient descent, deep neural networks, double descent curve)
  • Decision trees (e.g., random forests)
  • Model selection and validation (e.g., over- and underfitting, stratification, y-scrambling, learning curves, resampling, cross-validation)
  • Selected topics (e.g., learning theory, prediction error decomposition, predictive uncertainty, active learning)

Lernziele

The course introduces some of the basic concepts, problems, algorithms, and best practices in machine learning, including theoretical foundations, some application examples, and selected implementation aspects. The purpose of this course is to lay a principled foundation for analysis and prediction based on data, which will allow participants to successfully attend more advanced courses in data science and machine learning.

Weitere Details zum Inhalt der Vorlesung im ZEUS

Video Recordings

Die Livestreams und Vorlesungsaufzeichnungen finden Sie im dazugehörigen Kurs auf ILIAS.