Computational problems in multi-scale neuroscience

Résumé fourni par la source

Keynote address. Modern neuroscience increasingly involves integrating biological features across multiple scales of organization. While this approach opens exciting new vistas for studying the structure and function of the brain, it also prompts new methodological and computational challenges that cannot easily be met with conventional statistical methods. Here I will describe three such challenges and the solutions that have been developed. The first is the problem of spatial interpolation: how to generate dense maps of discretely sampled features, such as gene expression or intracranial electrophysiology, such that they can be made comparable to other intrinsically dense brain maps. Second is the problem of comparing spatially autocorrelated brain maps, and the diverse methods used to generate spatial autocorrelation-preserving randomized maps. The third is the problem of randomizing brain connectomes with continuous weights using optimization procedures such as simulated annealing. Collectively, these problems are increasingly understood and incorporated into workflows, laying the foundation for next-generation integrative neuroscience.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.