Decoder-style information map from latent classifier weights — dkge_info_map_from

Pulls a cross-fitted latent weight vector \(\beta^{(-s)}\) back to each subject's cluster space via \(m_s = A_s \beta^{(-s)}\), transports those maps to anchors, aggregates across subjects, and optionally performs inference.

Usage

dkge_info_map_from_classifier(
  fit,
  betas,
  renderer,
  lambda = 0,
  to_vox = TRUE,
  inference = c("none", "signflip", "parametric")
)

Arguments

fit: Fitted `dkge` object.
betas: Either a numeric vector of length `r` or a list of length `S` (number of subjects) containing one latent weight vector per subject.
renderer: Renderer produced by [dkge_build_renderer()].
lambda: Non-negative smoothing parameter applied via [dkge_anchor_aggregate()].
to_vox: Logical; when `TRUE` and a decoder is available in `renderer`, a dense voxel map is produced.
inference: One of `"none"`, `"signflip"`, or `"parametric"` specifying the group inference routine applied to anchor maps.

Value

A list containing the mean anchor field, optional voxel map, per-subject anchor maps, and inference outputs (t- and p-values when requested).

Examples

# \donttest{
toy <- dkge_sim_toy(
  factors = list(A = list(L = 2), B = list(L = 3)),
  active_terms = c("A", "B"), S = 6, P = 20, snr = 5
)
fit <- dkge(toy$B_list, toy$X_list, kernel = toy$K, rank = 2)
#> Warning: Argument 'kernel' is deprecated; use 'K' instead.
centroids <- lapply(toy$B_list, function(B) matrix(rnorm(ncol(B) * 3), ncol(B), 3))
renderer <- dkge_build_renderer(fit,
                                centroids = centroids,
                                anchor_xyz = matrix(rnorm(20 * 3), 20, 3),
                                anchor_n = 20,
                                anchor_method = "sample")
y <- factor(rep(c("class1", "class2"), length.out = length(fit$Btil)))
folds <- dkge_define_folds(fit, type = "custom",
                           assignments = list(c(1, 2), c(3, 4), c(5, 6)))
clf <- dkge_cv_train_latent_classifier(fit, y, folds = folds)
info <- dkge_info_map_from_classifier(fit, clf$beta_by_subject, renderer, to_vox = FALSE)
length(info$mean_anchor)
#> [1] 20
# }