Multicompartmental Modeling of Astrocytes

Astrocytes has a star-like structure with several branches connected to a soma. Until recently, experimental techniques had insufficient spatial resolution to reliably measure calcium activity at more than a few points. The advent of two-photon microscopy produced high-resolution imagery of fluorescence which measured a quantity related to calcium concentration. In this project, motivated by experimental measurements, we generalize the one-point model of Postnov (2007) to a multi-point model (one point per branch and one point in the soma). We compute time-dependent calcium concentration at these points by solving a system of coupled ODEs, and explore the interaction between processes and the soma. The research is performed by Evan Cresswell, Joel Tabak, Nathan Crock, and Gordon Erlebacher.

Feature Identification in Astrocytes

Video of Two-Photon Microscopy taken at high spatial resolution in an astrocyte within the cortex of a Ferret undergoing an experiment wherein patterns of different orientation are presented, reveals a complex pattern of activity that we would like to unravel. To this end, we apply machine learning techniques based on neural networks to identify the features that are most characteristic of the observed patterns. While we believe that these features are in part composed of diffusive and convective events, there may be additional fundamental characteristics to unravel. Rather than rely on human manual layer, we are investigating the use of variational Baysian autoencoders combined with auxiliary algorithms to allow for multimodal posterior and prior distribution with the goal of identify easily interpretable features that will help us understand the dynamics inherent to the astrocyte. The research is performed by >Nathan Crock, Gordon Erlebacher, and James Schummer.

Player Modeling in Procedural Games

Research is performed by Danial Smith, Fengfeng Ke, and Gordon Erlebacher

Simulated Annealing applied to Neural Networks

Research is performed by Brian Bartoldson, Gordon Erlebacher, and Adrian Barbu

Training of Robots in Challenging Terrain

Research is performed by Mario Harper, Emanuel Collins

Analysis of eye movement while searching imagies

Research is performed by Jjuan Llanos, Gordon Erlebacher, and John Jones