Solvers
SLCE — Supervised Linear Centroid-Encoder
Centroid projection-based adaptation method.
How It Works
Maps each sample toward its class centroid with an orthonormal linear projection (Ghosh & Kirby, Pattern Recognition 2024).
- Centroid Projection — projects features toward class centroids.
- Positive Eigenvalues — keeps only positive-eigenvalue directions (at most C-1).
- Ridge Regression — fits classifier in projected space.
Parameters
| Parameter | Default | Description |
|---|---|---|
| projection_dim | None | Max projection dimension (None auto-resolves to min(128, feature_dim, C-1)) |
from vyntri import Vyntri
from vyntri.data import split
s = split("./my_dataset", train=0.7, test=0.2, seed=42)
model = Vyntri(adaptation="slce")
model.fit(train=s.train, val=s.val)
# Diagnostics available
print(model.projection_.spectrum_)
print(model.projection_.n_positive_)Note: SLCE is not supported by the continual learning path. Use update() with adaptation="fk".