vyntri.adaptation

Projection & Shrinkage

Class / Function	Description
`FKProjection`	Discriminative whitening with eigenvalue flooring
`SLCEProjection`	Supervised Linear Centroid-Encoder projection
`AnalyticRidge`	Closed-form ridge regression with float64 solve
`shrink_within_class`	Shrinkage for within-class covariance (diagonal, ledoit_wolf)
`ledoit_wolf_shrinkage`	Data-estimated shrinkage intensity (Ledoit & Wolf, 2004)
`ledoit_wolf_covariance`	Shrinkage covariance estimator

Example

from vyntri import Vyntri
from vyntri.data import split

# Split your dataset
s = split("./my_dataset", train=0.7, test=0.2, seed=42)

# Adaptation via constructor (FK is default)
model = Vyntri(adaptation="slce")
model.fit(train=s.train, val=s.val)

# Shrinkage via constructor
model = Vyntri(shrinkage="ledoit_wolf")
model.fit(train=s.train, val=s.val)