The goal of this course to familiarize the student with fundamentalconcepts of machine learning in general and inductive rule learning inparticular. It will focus both on a predictive setting, where the goalis to learn a set of rules that collectively make a prediction, and adescriptive setting, where the goal is to learn a set of rules thatcollectively explain the data. The learning techniques will be firstillustrated for concept learning tasks in propositional logic, butlater also extended to first-order logic as well as to structuredprediction tasks.There are no prerequisites except for basic knowledge aboutalgorithms. The course is thus not only suited for computer sciencestudents but for all students that have a strong interest in machinelearning in data analysis.
Rules - the clearest, most explored and best understood form ofknowledge representation - are particularly important for data mining,because they offer the best tradeoff between human and machineunderstandability. This course will present algorithms for automatedrule learning and discovery as investigated in classical machinelearning and modern data mining. We will start with algorithms forlearning single rules in propositional logic, move on to learning rulesets with the covering or separate-and-conquer algorithm, inductivelogic programming algorithms for learning rules in first-order logic,and discuss approaches that allow to make predictions in structuredoutput spaces. Elementary data mining algorithms such as associationrule discovery will also be covered, as well as essential concepts ofmachine learning and data mining. It is thus suitable as a firstintroduction into these research areas.Most of the course will follow the book "Foundations of Rule Learning"that will appear in Springer-Verlag in early 2012.http://www.amazon.de/Rule-Learning-Essentials-Relational-Technologies/dp/3540751963
ECTS-Breakdown:--------------------------20 hours lectures53 hours preparation for exam.2 hours examination--------------------------75 hours: total--------------------------
The evaluation will depend on the examination at the end of term.