Formula: DP prior on community labels; Bernoulli edges
Usage
Dpsbm(adjacency, alpha = 1, n_iter = 30, seed = 42)
Arguments
- adjacency
An n x n symmetric 0/1 adjacency matrix; the
diagonal is ignored.
- alpha
Concentration, strictly positive.
- n_iter
Number of sweeps.
- seed
Seed of the deterministic stream.
Value
List with estimate (number of blocks), z,
counts, n_blocks, log_likelihood, n,
method.
Details
The block-to-block edge probabilities carry independent Beta(1,1)
priors and are integrated out, so a block pair (r, s) with e edges out
of t possible contributes B(1 + e, 1 + t - e) to the collapsed
likelihood. A node is then reassigned by the CRP prior times the
change in that likelihood, with the number of blocks learned rather
than fixed.
References
Kemp, Tenenbaum, Griffiths, Yamada & Ueda (2006),
AAAI-06:381-388.
Examples
Dpsbm(adjacency = 5L)
#> $estimate
#> [1] 1
#>
#> $z
#> [1] 0
#>
#> $counts
#> [1] 1
#>
#> $n_blocks
#> [1] 1
#>
#> $log_likelihood
#> [1] 0
#>
#> $n
#> [1] 1
#>
#> $method
#> [1] "Dirichlet-process stochastic block model"
#>
#> attr(,"class")
#> [1] "morie_rich_result" "list"
