Schemasdecomposition/tsne/v1

tSNEConfig

Variant

t-Distributed Stochastic Neighbor Embedding. See setup_tSNE.

Raw JSONv1Unknown properties rejected

Properties

k: integer≥ 1
Number of components to extract.
initial_dims: integer≥ 1
Initial dimensions.
perplexity: number≥ 0
Perplexity.
theta: number≥ 0≤ 1
Speed/accuracy trade-off.
check_duplicates: boolean
Check for duplicates.
pca: boolean
Perform an initial PCA step.
partial_pca: boolean
Use truncated PCA (irlba).
max_iter: integer≥ 1
Maximum number of iterations.
verbose: boolean
Print progress.
is_distance: boolean
Treat the input as a distance matrix.
Y_init: array | null≥ 1 items
Initial Y (embedding) matrix. NULL = random initialization. Must have one row per case.
pca_center: boolean
Center before the PCA step.
pca_scale: boolean
Scale before the PCA step.
normalize: boolean
Normalize the input.
stop_lying_iter: integer≥ 0
Iteration after which exaggeration stops.
mom_switch_iter: integer≥ 0
Iteration at which momentum switches.
momentum: number
Initial momentum.
final_momentum: number
Momentum used later in optimization.
eta: number
Learning rate.
exaggeration_factor: number
Early-exaggeration factor.
num_threads: integer≥ 0
Number of threads (0 = all cores).

Used by