Modul CS4337-KP12

Workload:

• 240 Hours private studies
• 90 Hours e-learning
• 30 Hours work on project

Contents of teaching:

• Foundations of Machine Learning and Data Science: Classification, regression, prediction: perceptrons, multi-layer perceptrons, and deep learning / Statistical principles: sampling, estimators, distribution, density, cumulative distribution, scales: nominal, ordinal, interval, and ratio scales, hypothesis testing, confidence intervals / Stochastic foundations, probabilities, Bayesian networks for the specification of discrete distributions, queries, query answering algorithms, learning procedures for Bayesian networks / Time series analysis: autoregression, integration, moving average (ARIMA), ordinal patterns, permutation entropy features, dynamic Bayesian networks and associated machine learning techniques / Inductive learning: version space, information theory, decision trees, rule learning / Ensemble methods, bagging, boosting, random forests / Automated machine learning / Clustering, k-means, analysis of variation (ANOVA), T-test, inter-cluster variation, intra-cluster variation, F-statistics, Bonferroni correction, MANOVA.
• Evolutionary Robotics: Biological basics of natural evolution / Evolutionary computation and optimization: coding, search spaces, genetic operators / Conducting evolutionary experiments with mobile robots in hardware and in simulation / Robot simulations and the reality gap / Concepts of reactive behavior and how to go beyond / Explanation of evolutionary dynamics in terms of nonlinear dynamics / Heuristic and empirical approach in robot experiments / Modular robotics for evolution of robot morphologies / Intensive discussion of state of the art methods, such as bridging the reality gap, novelty search, MAP elites, etc.
• Collective Robotics: Self-organization and feedback loops in systems / Basics of swarm behaviors, swarm robotics and behavior-based robotics / Robot swarms on land, water and in the air / Self-organized coordination of robots, autonomous assignment of tasks and roles, online distribution of tasks / Collective behaviors limited by local information, representative samples / Synchronization, estimate group size, mathematical modeling, micro-macro problem, random graphs / Collective decision making, urn models, opinion dynamics, speed vs accuracy tradeoff / Bio-hybrid robotics: animals and robots, plants and robots, cyborgs
• Machine Learning Lab: Methods and algorithms for the visualization, analysis and generation of medical image data, including current research work in the field of medical image processing / Basics of medical image processing – visualization and pre-processing of images / Image data augmentation techniques / Basics of connectionist networks in medical image processing / Convolutional networks and deep learning in medical image processing / U-Nets and generative adversarial networks (GANs) for the generation of medical image data / Generative models for medical image processing