Advantages and Pitfalls of Pattern Recognition

Horst Langer has developed methods for automatic alert systems and early warning on Mount Etna as well as tools that are routinely operated in the monitoring room of the institute and are part of the alert system for Civil Protection. Aside from his documented experience in the application of various pattern recognition techniques, he has also published computer programs for pattern recognition. Susanna Falsaperla has a long experience in the application of pattern recognition techniques and was among the first seismologists to apply automatic classification to seismic signals on volcanoes. She has made extensive use of pattern recognition in volcanology to relate multidisciplinary data to volcanic unrest and eruptive activity. Conny Hammer has worked on automatic classification of seismic signals in continuous data streams and has introduced novel concepts and tools into the seismological community from fields of machine learning (e.g., speech processing). Her automatic recognition tools are currently implemented in daily observatory routines. Besides automatic event detection, she has focused on the application of machine learning tools in seismic site characterization.

Part I: From Data to Methods1. Patterns, Objects, and Features2. Supervised Learning3. Unsupervised Learning

Part II: Example Applications4. Applications with Supervised Learning5.. Applications with Unsupervised Learning

Part III: A Posteriori Analysis6. What is a Failure? A-posteriori Analyses - Advantages and Pitfalls of Pattern Recognition Techniques7. Software Manuals