Institute of Cognitive Science
With my background in cognitive science and complex adaptive system, the main focus of my expertise lies in multi-agent simulations, reinforcement learning and bio-inspired algorithms, as well as the analysis of complexity, self-organization and emergence as properties of those systems.
Complexity, Self-Organization and Emergence in a Multi-Agent System through Microcosm Simulation
I am working on a multi-agent system simulating populations of simple organisms. Agents are mutually trained by their common experiences with actor-critic reinforcement learning in a curriculum fashion. The increasing complexity of the simulation motivates agents to adapt their behavior and interactions, developing a firm understanding of the dynamics in their environment. Communication among agents allows for cooperative behavioral strategies to emerge.
Complexity, self-organization and emergence are system properties that are often used in biology to analyze and describe complex systems in nature. In complex systems, due to the composition of many parts and the high degree of interconnectivity in their interactions, the prediction of the system's progression is inherently limited, especially for traditional reductionistic methodology. But their dynamics can also allow for the evolution of novel global solutions hidden in computational irreducibility.
Cortical Spike Synchrony as a Measure of Contour Uniformity
This project is based on a previous publication by Korndörfer et al. (2017) about Cortical spike synchrony as a measure of input familiarity by implementing long-range, horizontal, intercortical connections. In this project, the self-organization and synchronization of spiking networks using large-scale simulation of spiking systems (pulse-coupled excitatory spiking networks in a noisy environment) are investigated. By extending the previous project (Korndörfer, 2017) with retinotopy on the level of V1 and studying "gestalt principles" such as continuity and proximity.