Research
HH-SAE: Discovering and Steering Hierarchical Knowledge of Complex Manifolds
arXiv:2605.10536v1 Announce Type: cross Abstract: Rare semantic innovations in high-dimensional, mission-critical domains are often obscured by dense background contexts, a challenge we define as exti
arXiv:2605.10536v1 Announce Type: cross Abstract: Rare semantic innovations in high-dimensional, mission-critical domains are often obscured by dense background contexts, a challenge we define as extit{feature density conflict}. We introduce the extbf{Hybrid Hierarchical SAE (HH-SAE)} to resolve this by factorizing manifolds into a nested hierarchy of extbf{Contextual} (L_0), extbf{Atomic} (f_1), and extbf{Compository} (f_2) tiers. Evaluating across disparate manifolds, HH-SAE demonstrates superior resolution by extbf{``fracturing'' administrative clinical labels into physiological modes} and achieving a peak extbf{cross-domain zero-shot AUC of 0.9156 in fraud detection}. Path ablation confirms the architecture's structural necessity, revealing a 13.46% utility collapse when contextual subtraction is removed. Finally, knowledge-steered synthesis achieves a +9.9% AUPRC lift over state-of-the-art generators, proving that HH-SAE effectively prioritizes high-order mechanistic innovation over environmental proxies to enable high-precision discovery in high-stakes environments.
Source: arXiv cs.AI | 2026-05-12