PV056 Machine learning and data mining

Tue 22.5. 10-13, Tue 29.5. 11-14, Wed 13.6. 10-13, Thu 21.6. 9-12
Test 30 questions, max. 30 points
Machine learning
Inductive logic programming
Outlier and rare event detection
The process of knowledge discovery
Pre-processing for data mining
Association rule mining
Mining in various domains
+
Report
max. 10 points (>8: correct, with theoretical background; >5: correct, without a theory; >3: with minor errors;
on one of the following topics of your choice (to be changed)
Theory=deep understanding of one of the learning algorithm, for the first topic;deep description of (one of) a method, for the rest

Preliminary list (to be changed):
Data mining tools (RapidMiner or KNIME)
Regression trees
Learning from multi-instance data
Feature construction
Frequent patterns and association rule mining in first-order logic
Mining in geographic data (including logics for spatial data)
Privacy-preserving data mining

Course materials

* = is not subject to the final exam

Slides are available at IS MU -> PV056 -> Teaching materials.

Weka
(homepage) (datasets) (wiki) (how do I use WEKA's classes in my own code)
(multi-instance classification in weka) and (data sets)
Time series in Weka (two filters in weka) (TClass) (from wpi)
Artificial neural network plugins etc.
BioWeka

R
Home page of the R-project
R-project at CUNI

INTRODUCTION TO MACHINE LEARNING

Introduction
Classification. Version spaces (animation)
Learning decision trees. C4.5 (animation 1) (animation 2)
Evaluation of results
Bayesian learning. Instance-based learning. SVM.
Network models. Neural nets. Winnow
Inductive logic programming (multi-relational learning) (intro) (advanced) (aleph)
Unsupervised learning. Cluster analysis. CLUSTER. AutoClass
Outlier and rare event detection (intro) (KDD 2010 Tutorial)

ADVANCED MACHINE LEARNING

Ensemble methods. Bagging. Boosting. Random forests
Regression trees (C&RT)
Active learning
Propositionalization
Learning from streams (ICDM 2006 Tutorial)
Learning from multi-instance data
Semi-supervised learning
Preference learning
Meta-learning (intro) (Data Mining Advisor)

DATA MINING

Basics: KDD process. Data mining, dependence, association, interestingness. Tasks. CRISP-DM
Frequent patterns and association rule mining (intro)
and Apriori algorithm (example) (1) (2) (3)
First-order frequent patterns. Methods. RAP (here) (PhD. Thesis, auth.)
Data pre-processing (overview)
Predictive Model Markup Language
DBMiner Projecti description Tutorial and inductive query languages
Other data mining tools
Decision Support Systems
Introduction to Privacy-preserving data mining (pdf)
Swarm intelligence and data mining
    intro and swarm robotics (ppt)
    The bee algorithm and optimization (ppt) by Ziad Salem
    Ant colonies for data mining (ppt)
Distributed mining: decision tree learning*, frequent patterns mining (here) (PhD. Thesis, auth.)