Fundamental Methods for AI
Machine Learning for
Evolving Graph Data
Neuro-inspiredAI-Methods
Privacy-PreservingMachine Learning
Explanations for Trusted AI
Machine Learning for Evolving Graph Data (Prof. Rahm)
- Research is mostly restricted to static graphs, although most graphs change considerably over time
- We need to compute embeddings that encode the temporal information - but Representation Learning for temporal graphs is largely unexplored
- Goal:
- Integration and research on foundations for temporal graphs and representation learning within GRADOOP
- Build foundations of a graph streaming system that integrates methods for stream-based graph mining and learning
- Investigate incremental graph mining and learning techniques such as graph sketches, incremental grouping, incremental representation learning and incremental frequent pattern mining
- Application of graph stream analytics to be able to perform root cause analysis and anomaly detection
Neuromorphic Information processing for intelligent and cognitive data processing (Prof. Bogdan)
- Real Neuromorphic Information Processing requires models closer to the neurological dynamical system
- Integrate-and-Fire (IaF) neuron based Spiking Neural Networks (SNN) already outperform Deep Learning (DL) e.g. in recognition of handwritten patterns
- SNN still lacks of reliable and universal training algorithms dyn. changes in the trained network are not yet possible
- Synaptic plasticity is needed - not implemented in IaF neurons
- We envison dynamic synapses, based on the Modified Stochastic Synaptic Model (MSSM) (K. Ellaithy and M.Bogdan 2017)
Privacy-Preserving Machine Learning (Prof. Rahm)
- Goal: allow the analysis of person-related data with machine learning methods while guaranteeing a high degree of privacy such that the identity of individuals cannot be revealed
- Challenging when person-related data from multiple sources is combined, e.g., user profiles, patient data involving clinical and genomic data
- Requires to solve the problem of record linkage and a combination with ML-techniques
Argument-based Explanations for Trusted AI (Dr. habil. Baumann, Prof. Brewka)
- Explainability of AI systems: not just a desirable add-on, but a legal requirement.
- Strong recent focus in AI on computational models of argumentation:
- Identify arguments and their relationships in texts or formal knowledge bases
- Represent findings in argument graph; evaluate graph to identify reasonable argument sets
- Accept their conclusions
- Goal: exploit progress in the field to generate convincing explanations; options:
- Translate representation formalism used to argument graph and benefit from the graph’s explanation facilities, or
- Install an argumentation framework as a monitoring system for an underlying AI system
