5 papers
What Makes a Representational Prior Work? Feature Families, Label-Free Invariances, and Critical Windows in Grokking
Gunner Levi Howe
Companion work showed the grokking delay is causally the time to form task-structured representations, injectable via a contrastive prior. Here we characterize what makes such a pr…
Structure-Specific Representational Priors Causally Control the Grokking Delay
Gunner Levi Howe
Grokking -- generalization long after training-set interpolation -- has been accelerated by structure-agnostic interventions (gradient filtering, weight-norm clamping, geometric pe…
Intrinsic-Noise Consolidation: A Doob-Barrier-Conditioned Diffusion Turns Analog Device Noise into a Continual-Learning Resource
Gunner Levi Howe
On analog neuromorphic hardware, intrinsic device noise is normally an accuracy tax. We ask whether it can instead consolidate memories. We cast per-synapse consolidation as a Doob…
Level-Crossing Density as a Mesh-Free High-Frequency Auxiliary Loss for Implicit Neural Representations
Gunner Levi Howe
The Minkowski functionals of a field's excursion sets -- area, boundary measure, and Euler characteristic -- describe its level-set morphology; the Euler characteristic is the chea…
Heckman-Corrected Epistemic Uncertainty: Selection on Unobservables Defeats Importance Weighting
Gunner Levi Howe
Training data for machine learning is routinely collected by a selection process the model never sees: loans are observed only when granted, outcomes only when a test was ordered.…