Data Fields | |
vl_aib_node * | nodes |
vl_aib_node | nentries |
vl_double * | beta |
vl_aib_node * | bidx |
vl_aib_node * | which |
vl_aib_node | nwhich |
vl_aib_prob * | Pcx |
vl_aib_prob * | Px |
vl_aib_prob * | Pc |
vl_uint | nvalues |
vl_uint | nlabels |
For internal use only.
The implementation is quite straightforward, but the way feature values are handled in order to support efficiently joins, deletions and re-arrangement needs to be explaiend. This is achieved by a layer of indrirection:
Px
[i] actually refers to the node nodes
[i]. Entires to nodes
Total number of entries (= # active nodes)
Minimum distance to an entry
Closest entry
List of entries to update
Number of entries to update
Joint probability table
Marginal.
Marginal.
Number of feature values
Number of labels