The Bernstein Center Freiburg

Informal Seminar
Benjamin Dunn

Kavli Institue for Systems Neuroscience
Norwegian University of Science and Technology

Incorporating the effects of the unknown
in the analysis of neural data

With recent advances in high-throughput recordings, researchers are turning to statistical models to interpret the data. These methods, however, are limited by the extent to which the population is covered. The effect of what remains hidden on what can be inferred is currently not understood and poses a significant challenge. Here we have sought for ways to understand and correct for the effects of subsampling in inference for the paradigmatic case of the kinetic Ising model. We show that these effects provide two sources of errors. The first is a systematic bias that can be accounted for directly within the mean field solution. The second term involves correlations between the visible and hidden populations directly and can be shown to be small for systems with weak couplings. We also derive a second method to account for these errors by including hidden nodes and using approximate methods to marginalize out their effects. Through application of these methods on Ising networks of varying relative population size and coupling strength, we assess how these unknown variables can influence inference and to what degree they can be accounted for.

Tuesday, February 11, 2014

14:00 h
BCF Library
First Floor
Hansastr. 9a
The talk is open to the public. Guests are cordially invited!