Lectures

Science is the belief in the ignorance of experts.

I have taught Artificial Intelligence, Machine Learning, Knowledge Discovery and Temporal Data Mining continuously since 2019 in bachelor’s and master’s programmes. My courses combine theoretical foundations with reproducible implementations in R, version-controlled software, case studies and project-based learning. They address not only how algorithms work, but also the assumptions under which they are valid, how their performance should be evaluated, when they can fail, and how uncertainty and limitations should be communicated.

A recurring objective is to connect statistical and machine-learning methods with understandable knowledge and practical decision support. Course materials therefore integrate data visualization, clustering, classification, time-series analysis, explainable AI, human-in-the-loop methods, self-organization and interdisciplinary applications. The archived materials below document selected courses, lectures and conference presentations.

Note: Materials are provided for academic teaching and personal study. Please cite the corresponding publications or software packages when reusing methodological content.

Knowledge Discovery for Heterogeneous Data II:

Chapter 4: Multivariate Structures in Data

• Methods of dimensionality reduction
• Graph theory with applications of neighborhood graphs
• Focusing projection methods
• Quality assesment is based on Graph theory

Chapter 5: Unsupervised Machine Learning

• Emergent self-organizing maps and the U-matrix
• Conventional cluster analysis
• Combining dimensionality reduction with cluster analysis
• Quality measurement and Benchmarking of Algorithms

Chapter 6: Classification

• Subsymbolic classifiers: KNN and Naiv Bayes
• Subsymbolic classifiers: Support Vector Machines and Neural Networks
• Estimation of generalization ability
• Benchmarking

Chapter 7: Knowledge

• Knowledge and Knowledge Bases
• Understandability and Post-hoc-Explainers of subsymbolic classifiers
• Symbolic Classifiers
• Building data-driven explainable Artificial Intelligence Systemss

Knowledge Discovery for Heterogeneous Data I:

Chapter 1: Pattern of Thought in Knowledge Discovery

• The circular process of KD
• The difference to statistics

Chapter 2: Basic Exploration

• Scales, Types of Errors in Measurement
• Statistical Descriptions of Variables

Chapter 3: Univariate Structures in Data

• Transformations
• Bayesian Theorem and its Applications
• Multimodal Distributions and Gaussian Mixture Models
• Skewed Distributions
• Relative Differences

Chapter 4: Multivariate Structures in Data

• Handling Missing Values
• Correlations and Distances
• Challenges of High-dimensionality

Databionic Methods for Artificial Intelligence:

Databionics means the transfer of algorithms for data processing from nature. One major part of databionicsis artificial intelligence (AI). In this context, AI is restricted to seeking, explaining and emulating intelligentbehavior in the form of a computational processes.

Chapter 1: Databionics and Randomness

• Introduction into Databionics
• Chance - The Prerequesite for Emergence

Chapter 2: Introduction into Supervised Artificial Neural Networks

• Perceptron Algorithm
• Multilayer Perceptron Network with Backpropagation
• Building a Classifier with MLP-BP
• Outlining Pitfalls and Challenges using ML Theory
• Application: Forecasting with Supervised Artificial Neural Networks

Chapter 3: Unsupervised Learning Focused on Neural Networks and Emergence

• Types of Self-Organizing Maps (SOM)
• Excurs: Exploring the Brain Using Neuroscience
• Clustering with Online Self-Organizing Maps (SOM)
• Projection and Visualization with Emergent SOM (ESOM)
• The Curse of Dimensionality in High-Dimensional Data and Quality Assesment Of Projections
• Simplified ESOM and Generalized U-matrix for Projections
• Emergence Theory

Chapter 4: Behavior-Based Systems

• Artificial Life and Conway's Game of Life
• Principles of Collective Behavior and Swarm Intelligencen
• Agents, Self-Organization and Schelling's Model
• Ant-based Systems: Ant Colony Optimization and Ant-based Clustering
• Particle Swarm Optimization
• Exploiting Swarm Intelligence using of Game Theory
• Databionic Swarm: Swarm-based Projection and Cluster Analysis of Complex Use Cases

Chapter 5: Evolutionary (EA) and Genetic Algorithms (GA)

• Biological Foundations
• Strategies for EA
• Mutation & Recombination in the Context of GA
• Applying GA and EA to Optimization Problems
• Distribution Optimization: Outperforming Expectation Maximization Algorithm in Case Of Gaussian Mixtures

Chapter 6: Knowledge Discovery in Genes and Their Products

• From DNA to Proteins: Process and Measurement Approaches
• Methods for Analyzing Gen Expressions
• Excurs: Gen Self-Regulation and Epigenetics
• Introduction of Biological Databases and Ontologies (Gene Ontology)
• Overrepresentation Analysis and Functional Abstraction
• Excurs: Exploiting a Concept of Information Retrieval for the Clustering of Genes

Temporal Data Mining:

Chapter 1:

• Time is Relative

Chapter 2:

• Decomposition Model

Chapter 3:

• Time Series as Stochastic Process
• Box Jenkins Methodology

Chapter 4:

• Approaches for Calculating Differences
• Introducing Risk Measures: Volatility and Beta
• ARCH/GARCH

Chapter 5:

• Introduction into Time Series Forecasting
• Forecasting with Decomposition Model of Facebook's Prophet
• Forecasting Evaluation Using Machine Learning Theory
• Forecasting with Machine Learning Ensemble

Chapter 6:

• Markov Chains
• Hidden Markov Model
• Explaining Baum-Welch and Viterbi Algorithm on the Example of Sunspots

Chapter 7:

• Introduction into Fourier Analysis
• Applying Fourier Analysis to Sunspots
• Windowing Effect
• DFT Spectrum of Common Cases
• Filtering with DFT
• Low-pass Fourier Filtering on the example of High-Frequency Hydrobiologial Data
• Short-Time Fourier Analysis and Heisenberg's Uncertainty
• Gabor Transformation

Chapter 8:

• Introduction into Wavelet Analysis
• Comparison of Time Series Representations to Wavelet Analysis

Chapter 9:

• Introduction into Time Series Classification
• Swarm-based Cluster Analysis using Dynamic Time Warping

Chapter 10:

• Temporal Knowledge Discovery by Non-Temporal Inference
• Temporal Inference: Pattern Evolution, Episodes, Sequential Pattern Mining and Association Rules
• Temporal Reasoning: From Allen's Interval Logic to Unification-based Temporal Grammars

Supplementary Introduction into R

Following are some of the lectures and talks I gave during my teachings under the tutelage of Prof. A. Ultsch or in conferences. More general lectures on data science:

• Approaches to Distribution Analysis
• Temporal Knowledge Discovery in Big Data Time Series
• Models of Income Distributions for Knowledge Discovery
• Introduction in the Concept of Swarms

Cluster Analysis and Visualization of such results:

• Projection based Clustering
• Knowledge discovery from low-frequency stream nitrate concentrations: hydrology and biology contributions
• Cluster Analysis of the World Gross-Domestic Product Based on the Emergent Self-Organization of a Swarm
• DataBionicSwarm (DBS)
• Visualization and 3D Printing of Multivariate Data of Biomarkers
• Benchmarking Cluster Analysis Methods using PDE-Optimized Violin Plots

Methods of Dimensionality Reduction and their Evaluation approaches:

• Investigating Quality Measures of Projections for the Evaluation of Distance and Density-based Structures of High-Dimensional Data
• Quality Measures of Projections
• Neighbor Retrieval Visualizer - NeRV

Two brief summaries on knowledge discovery and temporal data mining and basics about high-dimensional data in German:

• Summary - Knowledge Discovery (KD)
• Summary - Temporal Data Mining and KD
• Hochdimensionale Daten