concept
active
concept:pullback-geometry-behavior-aware-metricPullback Geometry (Behavior-Aware Metric)
Neighborhood — ranked by edge-count
Communities (1)
community
- Neural Geometrymembers_of
Concepts (1)
concept
- Manifold Steeringassociated_withCentral framework: steering neural networks by intervening along the curved manifold where a concept lives, rather than in straight lines through activation space.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- The method of optimizing steering interventions in activation space to produce outputs that follow the behavior manifold, independent of the representation manifold.
- The bidirectional correspondence M_h ↔ M_y, indicating that geometry in representation is not incidental but causally shapes behavior.
- Geometry-behavior correlate robust to pooling strategy, distance metric, and frozen encoderfinding0.734Robustness checks confirm sign stability.
- Core finding: the structure models use internally (representations) is precisely reflected in their external behavior (outputs).
- Central question: does geometry in activation space causally determine behavior?
- Paradigm of finding the right geometry (manifold) for principled control.
- Relationship where geometry summaries predict behavioral thresholds.