Python based Machine Learning with Automotive Applications

Description

The topic of Artificial Intelligence (AI) is currently becoming more and more important, in particular in areas where processes are automated and many data are processed. Especially in automotive area as well in the virtual development process as in the field of testing, numerous applications are conceivable in this context. A part of artificial intelligence is machine learning, which is becoming increasingly important in addition to classical rule-based expert systems. This current development is due to the generation of ever-larger datasets (big data) as well as more powerful computers for their processing. Especially in the automotive environment, extensive data are generated in the context of simulation or testing, for which an automated analysis is often sought. In addition to the classical interpretation of individual simulation or testing results, the methods of machine learning allow a new view at models and results. Based on the analysis of numerous results (big data), e.g. from parameter studies, it is possible to derive Artificial Intelligence using methods of machine learning, which is then used to evaluate further simulations or tests. Python is currently the most popular programming language for data analysis and machine learning. The freely available Python library Scikit-Learn provides a user-friendly entry to the relevant procedures. Especially the application of artificial neural networks (Deep Learning) has become very popular lately. The software TensorFlow developed by Google and the Python library Keras based on it provide a beginner-friendly access.

Who should attend?

The seminar addresses participants coming from CAE or testing field who want to take the first steps in machine learning based on their Python knowledge. It is assumed that basic Python knowledge - e.g. as it is conveyed in the carhs.training seminar Introduction to the Python Programming Language of the same trainer - exists.

Course Objectives

The seminar gives an introduction to machine learning based on the programming language Python. This includes, as a start, topics of data analysis, preparation and visualization. In the second step, methods of machine learning are studied using the Python packages Scikit-Learn and Keras or TensorFlow. Practical exercises will deepen the topics discussed and discuss possible applications in CAE or testing. An important aspect of data analysis is the extraction of features from CAE or testing data for the use in machine learning. After the seminar participants will be able to tackle the implementation of their own tasks. This also includes evaluating various methods of machine learning regarding their applicability to one?s own tasks and to deepen the methods based on the discussed Python packages.

Course Contents

• Basics of data analysis with Python
  • Data structures
  • Concepts of data preparation
  • Extraction of features for machine learning methods
  • Data visualization
  • The Python packages Numpy, Scipy, Pandas, Matplotlib
• Machine Learning with Python
  • Methods for classification and regression analysis
  • The Python Package Scikit-Learn
  • Deep Learning and Neural Networks with Keras, TensorFlow
• Applications motivated by CAE or testing background
  • Introductory examples
  • Discussion of possible deeper applications
  • Procedure for implementing your own ideas

Dr. André Backes
TECOSIM Technische Simulation GmbH

Dr. André Backes studied Mathematics at the University of Duisburg. From 2000 to 2006 he was a researcher at the Institute for Mathematics at the Humboldt University in Berlin. His PhD studies at the chair for Numerical Mathematics introduced him to the field of CAE. Since 2006 he works at TECOSIM GmbH and among other topics specialized in NVH. In the area of Virtual Benchmarking he helped developing the TECOSIM-owned process TEC|BENCH where also the Python language was used. In current research projects he investigates the use of Python-based methods for data analysis and machine learning in the CAE process. Since 2020 he has been working at TECOSIM Stuttgart.

This course is available as an in-house seminar. The trainer comes to your site - you save travel expenses and travel time. Get an offer.

Your contact person

Dr. Dirk Ulrich
Phone.: 06023 - 96 40 - 66
E-Mail: dirk.ulrich@carhs.de

Request now!