Content and procedure
Machine Learning is becoming an elementary value and success factor in digitization. This workshop provides an overview of the possibilities, methods and application areas of machine learning. This kickstart consists of the two areas Supervised and Unsupervised Machine Learning.
Topics include classification methods, Bayes classifiers, linear classifiers, decision trees and neighboring classifiers. Not only the technical aspects of the methods are explained, but also the respective advantages and disadvantages as well as exemplary application cases. Finally, an insight into neural networks and deep learning will be provided. The workshop is continuously accompanied by practical exercises.
After completing the workshop the participants are familiar with modern Supervised and Unsupervised Machine Learning approaches.
Our experienced instructors have several years of professional experience in various industries. Thus we guarantee a high practical relevance of our trainings and seminars.
- Speaker: Prof. Peer Kröger & Maria Egorova
- Language: English
- 19 March 2019
- 10:00 – 17:15
- Sofitel Hotel Munich Bayerpost, Bayerstraße 12, 80335 Munich, Germany
Insight into different relevant methods of Machine Learning
Knowledge of advantages and disadvantages for certain applications
Kickstart to use Machine Learning in your daily work routine
Prof. Dr. Peer Kröger
has been Professor of Data Science at the Chair of Data Systems and Data Mining at Ludwig Maximilian University (LMU) Munich since 2015. In 2013, he founded the Data Science Lab at the LMU with colleagues and has been its director since then. He has been active in science and business for many years as a consultant and trainer in the fields of data management, data analytics and data science, e.g. as a board member of the Archaeo-Bio-Center of the LMU, as a member of the steering committee of the Data Science Certificate Program of the LMU, and as managing director of Applied Data Science GmbH.
- Introduction to Machine Learning and why we need it: descriptive analytics vs. predictive analytics
- Classification of Machine Learning in the Business Context: Application Examples
- Clarification of the most important basic concepts and definition of the problems in Supervised Learning
- Bayesian classifiers
- Main principle of Bayesian classifiers (theorem of Bayes, Baye’s decision rule)
- Example implementation of the basic idea: naive Bayes classifiers
- Discussion of the advantages and disadvantages on the basis of exemplary applications
- Linear classifiers
- Functionality of linear classifiers
- Illustration of the basic idea: Support Vector Machines (SVMs)
- Extension of SVMs to non-linear classification (Kernel Machines)
- Discussion of the advantages and disadvantages on the basis of possible use cases
- Decision trees
- Basic principle of decision trees
- Example implementation of the basic idea: discussion of different split strategies
- Extension of decision trees: Pruning to avoid overfitting
- Discussion of the advantages and disadvantages with the help of real-life use cases
- Next neighbours (NN) classifiers:
- Basics of NN classifiers
- Example implementation of the basic idea: discussion of different weightings and voting strategies
- Discussion of the advantages and disadvantages of the application using the example of use cases
- Evaluation of the quality of learned models (classifiers)
- Discussion of the challenges, in particular of overfitting
- Overview of measures against overfitting: the train-and-test paradigm
- Basic introduction to the concepts of neural networks and Deep Learning
- Introduction to Data Mining and Unsupervised Learning and why we need it: labeled vs. unlabeled Data
- Classification of Unsupervised Learning in the Business Context: Application Examples
Basic knowledge in statistics, mathematics and computer science is necessary. A laptop is required for the practical exercises.