eXplainable Artificial Intelligence

Pointwise-in-Time Diagnostics for Reinforcement Learning during Training and Runtime
A framework for explaining RL agents during training and runtime based the definition of an agents desired behavior through linear temporal logic (LTL).
Noel Brindise, Andres Felipe Posada-Moreno, Cedric Langbort, Sebastian Trimpe
Cite URL
ECLAD: Extracting Concepts with Local Aggregated Descriptors
A novel method for automatic concept extraction and localization based on representations obtained through pixel-wise aggregations of CNN activation maps.
Andres Felipe Posada-Moreno, Nikita Surya, Sebastian Trimpe
Cite Project Project DOI URL
Scale-Preserving Automatic Concept Extraction (SPACE)
We introduce the Scale-Preserving Automatic Concept Extraction (SPACE) algorithm, as a state-of-the-art alternative concept extraction technique for CNNs, focused on industrial applications.
Andres Felipe Posada-Moreno, Lukas Kreisköther, Tassilo Glander, Sebastian Trimpe
Cite Project Project DOI URL
Scalable Concept Extraction in Industry 4.0
Applying concept extraction (ECLAD) for explaining CNNs in industrial use cases.
Andres Felipe Posada-Moreno, Kai Müller, Florian Brillowski, Friedrich Solowjow, Thomas Gries, Sebastian Trimpe
Cite Project Project DOI
Explaining Neural Networks Through Concept Extraction
This research project focuses on the methodological and application aspects of how to explain what a deep neural network learns through training. Specifically, this project tackles the development of methods for concept extraction, localization, and learning, as well as their applications in industrial scenarios.
Explaining Neural Networks Through Concept Extraction