Spin-glass order-parameter diagnostics for latent phase structure

methods / Theoretical / Analytical / Geometric analysis / Spin-glass order-parameter diagnostics for latent phase structure

Spin-glass order-parameter diagnostics for latent phase structure

Techniqueadvanced

Imports diagnostics from spin-glass statistical physics (overlap distributions between sampled latent codes, order parameters, susceptibility) to classify a real trained model's latent space into ordered, disordered, or glassy phases, rather than reporting a single smooth curvature or dimension number.

Used in (1 observation)

Real trained VAE latent spaces organize into measurably distinct ordered, disordered, and glassy phases, and shifting toward a more ordered phase improves anomaly detectionmeasured

structure: Curvature profile of the representation manifold · models: VAE trained on Mars Rover Mastcam imagery (Ascarate, Lebrat, Santa Cruz, Fookes & Salvado), VAE trained on Galaxy Zoo 64 imagery (Ascarate, Lebrat, Santa Cruz, Fookes & Salvado) · paper: High-Dimensional Latents Should Be Diagnosed Through Phase Structure