Maier, Alexander: Identification of timed behavior models for diagnosis in production systems. 2015
Content
- Introduction
- Motivation
- Relevance to Related Work
- Contribution of this Thesis
- Realized and Potential Application Scenarios
- Overview
- Part I Foundations
- The Core Issue: Modeling and Learning Time
- Formalisms
- Complexity of Identification of Finite Automata
- The Identification Problem
- Identification frameworks
- Complexity of Identification of Finite Automata
- Identification of Automata and Model-Based Anomaly Detection
- Part II Algorithms and Theory
- Algorithmic Results
- Requirements on modeling formalism and identification algorithm
- Bottom Up Timing Learning Algorithm (BUTLA)
- Online Timed Automaton Learning Algorithm (OTALA)
- Anomaly Detection
- Adaptive Learning
- Theoretical Results
- Runtime Analysis of the Identification Algorithms
- Evaluation of the Learning Error
- Convergence and Identification In The Limit
- Top-Down vs. Bottom-Up
- Splitting vs. Non-Splitting
- Online vs. Offline Identification
- Runtime analysis of the anomaly detection algorithm
- Empirical Results
- Part III Applications
- Part IV Conclusion