Joint Shared/Specific Similarity Clustering Wrapper

Usage

mrf3_specific_clustering(
  recon,
  shared_specific,
  shared_mode = c("average", "ao_reg"),
  shared_dat_use = "ALL",
  shared_k = NULL,
  shared_method = c("Spectral", "PAM"),
  shared_similarity_type = c("second", "first"),
  specific_k = NULL,
  specific_method = c("Spectral", "PAM", "Proximity", "Enhanced_Proximity"),
  specific_prox_method_cl = c("PAM", "Spectral"),
  specific_similarity_type = c("second", "first"),
  tune_method = "silhouette",
  gap_w = "uniform",
  gamma = 0.1,
  alpha_init = NULL,
  ao_max_iter = 20,
  ao_tol = 1e-04,
  knn_q = NULL,
  hollow = TRUE,
  ao_symm = TRUE,
  ao_verbose = FALSE,
  ...
)

Arguments

recon: Reconstruction output from get_reconstr_matrix().
shared_specific: Output from get_shared_specific_weights().
shared_mode: Shared clustering mode passed to cluster_shared_similarity().
shared_dat_use: Shared weight source for "average" mode.
shared_k: Optional k for shared clustering. Required for "ao_reg" mode.
shared_method: Clustering backend for shared "average" mode.
shared_similarity_type: Similarity structure for shared "average" mode.
specific_k: Optional k configuration passed to cluster_specific_similarity().
specific_method: Clustering backend for specific clustering. Choose from "Spectral", "PAM", "Proximity", and "Enhanced_Proximity".
specific_prox_method_cl: Proximity clustering backend used when specific_method is "Proximity" or "Enhanced_Proximity".
specific_similarity_type: Similarity structure for specific clustering.
tune_method: Tuning criterion for PAM when tuning k.
gap_w: Weighting scheme for spectral eigengap when tuning k.
gamma: L2 regularization for shared AO fusion.
alpha_init: Optional initial AO fusion weights.
ao_max_iter: Maximum AO iterations.
ao_tol: AO convergence tolerance.
knn_q: Optional kNN sparsification for shared AO fusion.
hollow: Logical; whether to zero diagonal in shared AO fusion.
ao_symm: Logical; whether to symmetrize kNN sparsification in AO fusion.
ao_verbose: Logical; whether to print AO diagnostics.
...: Additional arguments passed to clustering backends.

Value

A list with shared and specific clustering outputs.