Hilbert States for Decision-Making under Partial Observability
Organizations: University of Bristol, United Kingdom
Abstract
Decision-making under partial observability can be fundamentally limited by information, even with known dynamics: distinct physical states can induce the same observation history while favoring different actions. We study three questions: when is full-state optimality unattainable, what information exactly preserves the best observation-based decision, and how does finite-data learning affect reliability? First, we characterize decision ambiguity through conditional action preferences. We give a zero-regret criterion, a positive lower bound for conflicting uniquely optimal actions on positive-probability regions with uniformly positive cost gaps, and an exact binary-action regret formula. This also shows why physical-state ambiguity alone need not cause decision loss. Second, for two actions and squared terminal-goal costs in the plane, we derive an exact inner-product representation of conditional expected cost differences using three conditional moment contrasts and a goal-dependent feature. This Hilbert state preserves observation-optimal actions without hidden-state reconstruction. Under nondegenerate history and goal variation, its three-dimensional form is minimal among exact bilinear cost-contrast representations. Third, we relate finite-data state errors to excess decision regret via goal-projected errors, action margins, and goal geometry. Physics-based simulations verify information-limited regret, exact action preservation, and the role of second moments. Held-out real-SAR experiments demonstrate feasible goal-conditioned decisions from masked images, while finite-data evaluations reveal goal-dependent reliability. Together, these results separate irreducible information loss, exact representation sufficiency, and learning-induced excess regret.
Figures & tables
| Quantity | Dev | Shifted dev |
|---|---|---|
| Max. identity discrepancy | ||
| Hilbert action agreement (%) | 100.00 | 100.00 |
| Ranks of | ||
| Mean-only disagreement (%) | 0.61 | 0.49 |
| Mean-only excess regret | 0.00280 | 0.00206 |
| Hilbert representation regret | 0.00000 | 0.00000 |
| Quantity | Dev | Shifted dev |
|---|---|---|
| Observation ambiguity (%) | 50.00 | 50.00 |
| Action conflict (%) | 13.49 | 15.13 |
| Conflict given ambiguity (%) | 26.98 | 30.27 |
| Full-state optimal risk | 324.9373 | 394.6059 |
| Observation-optimal risk | 325.2692 | 395.0036 |
| Irreducible regret | 0.3319 | 0.3977 |
| Estimator | DEV | SHIFTED_DEV |
|---|---|---|
| Mean-only ( ) | 0.09209 | 0.08186 |
| Full moment ( ) | 0.12800 | 0.16057 |
| CV shrinkage | 0.08014 | 0.09518 |
| Goal family | Mean | Full | CV |
|---|---|---|---|
| DEV: extrapolated | 0.13680 | 0.13807 | 0.08931 |
| Shifted: novel direction | 0.06181 | 0.20998 | 0.08709 |
| Method | Regret (mean std.) | Group macro |
|---|---|---|
| Constant state | 0.04364 | |
| Mean-only | 0.03347 | |
| Direct contrast | 0.02979 | |
| Hilbert-3D | 0.03215 |
| Regime | Columns | Rows |
|---|---|---|
| Center vertical, 16 | ||
| Left vertical, 16 | ||
| Right vertical, 16 | ||
| Horizontal, 16 | ||
| Center vertical, 24 |
| Observation regime | Constant state | Hilbert-3D | Mean-only | Direct contrast |
|---|---|---|---|---|
| Center vertical, 16 | 0.05997 | 0.04025 | 0.04296 | 0.03694 |
| Right vertical, 16 | 0.05997 | 0.05611 | 0.06155 | 0.06490 |
| Left vertical, 16 | 0.05997 | 0.08501 | 0.06494 | 0.09487 |
| Horizontal, 16 | 0.05997 | 0.07563 | 0.06347 | 0.08440 |
| Center vertical, 24 | 0.05997 | 0.04391 | 0.04779 | 0.04167 |
Decision-Making under Uncertainty
Decision-making under uncertainty studies how to choose actions when outcomes, states, or models are imperfectly known.
See all 213 papers