Methods for measuring and generation of whitebox models of human behavior
Subprojects:
- Subproject SP1.1: Optimization of behavioral experiments with AI
Lead: Frank Jäkel, Participants: Kristian Kersting, Stefan Roth, Ralf Galuske - Subproject SP1.2: Inverse Reinforcement Learning explains human behavior
Lead: Jan Peters, Participants: Constantin Rothkopf, Heinz Koeppl, André Seyfarth - Subproject SP1.N: Development of new methods for investigating human problem-solving behavior
Lead: Constantin Rothkopf, Participant: Frank Jäkel
Methods to analyze blackbox models in machine learning
Subprojects:
- Subproject SP2.1: Probabilistic models of and with deep neural nets
Lead: Stefan Roth, Participants: Kristian Kersting, Heinz Koeppl - Subproject SP2.2: Inverse Reinforcement Learning explains machine behavior
Lead: Heinz Koeppl, Participants: Constantin Rothkopf, Jan Peters - Subproject SP2.3: Hybrid models: deep and probabilistic, graphical models interact mutually
Lead: Constantin Rothkopf, Participants: Frank Jäkel, Kristian Kersting
Transparent cooperation between human and machine using whitebox models for both
Subprojects:
- Subproject SP3.1: Human and machine walking together
Lead: André Seyfarth, Participants: Jan Peters, Ruth Stock-Homburg - Subproject SP3.2: Humans and machines learning together
Lead: Kristian Kersting, Participants: Constantin Rothkopf, Frank Jäkel - Subproject SP3.3: Human trust in self-explaining AI
Lead: Ruth Stock-Homburg, Participants: Jan Peters, Kristian Kersting, Ralf Galuske - Subproject SP3.N: Using human behavior for automated explanations
Lead: Frank Jäkel, Participant: Constantin Rothkopf
