Skip to contents

Rank-weight spec: TDNN asymptotic (Demirkaya two-scale DNN, k = 1)

Source: R/wrank-tdnn-wrapper.R
rank_weight_tdnn_asymp.Rd

Returns a rank_weight spec whose eval_batch uses the asymptotic TDNN rank table from tdnn_rank_table_asymp(). Structurally identical to rank_weight_tdnn_exact() but uses the Beta order-statistic approximation, which gives full rank support (all n entries are non-zero).

Usage

rank_weight_tdnn_asymp(s1, s2, dim, normalize = TRUE)

Arguments

s1

Positive integer; smaller U-statistic order (s1 < s2).

s2

Positive integer; larger U-statistic order.

dim

Positive integer; localization dimension.

normalize

Logical; renormalize the rank table to sum 1 (default TRUE; guards floating-point drift).

Value

A rank_weight spec (list with class "rank_weight").

See also

rank_weight_tdnn_exact(), tdnn_rank_table_asymp(), weights_tdnn_asymp()

Examples

rw <- rank_weight_tdnn_asymp(s1 = 3L, s2 = 6L, dim = 2L)
R <- matrix(c(2L, 1L, 3L, 4L, 5L, 6L,
              6L, 2L, 1L, 3L, 4L, 5L), nrow = 6)
W <- rank_weight_eval_batch(rw, R)
colSums(W)  # c(1, 1)
#> [1] 1 1