search

open-connectome-classes / StatConn-Spring-2015-Info

introductory material
18 stars 4 forks source link

higher-dimensional SBMs #261

Open mblohr opened 9 years ago

mblohr commented 9 years ago

Is there ever a need to model a higher-dimensional (>2 dimensional) SBM? If so, is there an example?

akim1 commented 9 years ago

Higher-dimensional as in spatial dimensions?

mblohr commented 9 years ago

In spatial dimensions, i.e., an SBM that's say 3-dimensional. Is there a scenario in which that extra dimension would be useful, e.g., dynamics?

akim1 commented 9 years ago

Oh. I would think a neuron is less likely to be connected to other neurons that are further away and more likely to be connected to those are closer.

ElanHR commented 9 years ago

Assuming you have spacial locations of/distances between verticies, you could easily modify SBM so that the probability of an edge is dependent on not only on the block membership of the two involved vertices but also the distance between them. Alternatively, if you only care about a first order approximation of this effect you could likely just work this into the inter-block parameters (blocks that are close to one another are more likely to be connected).

jtmatterer commented 9 years ago

@mblohr I don't understand what you mean by "spacial dimension" with regards to a SBM. Could you please elaborate, maybe describe what the two dimensions of the SBM we've been using correspond to?