Spectral_Decay

Higher is Better

[0, 1]

Description

Rate of eigenvalue decay indicating information concentration in leading dimensions

0.6 * sigmoid(-slope_log_eigenvalues) + 0.4 * (λ₁ / Σλᵢ)

Higher = faster exponential decay = information concentrated in few dimensions. Indicates dimensionality efficiency.

Other metrics in the Latent Space category

Multi-method dimensionality efficiency score combining variance thresholds, Kaiser criterion, elbow detection, and spectral decay

Effective dimensionality measure from eigenvalue distribution (trajectory: lower is better, steady-state: higher is better)

Multi-method anisotropy combining log-ellipticity, condition numbers, ratio variance, entropy, dominance, and effective dimensionality