YUKTI: From Natural-Language Situations to Robust, Verifiable Decisions An Uncertainty-Typed Proposition IR, Assumption-Robust Pareto Frontiers, and a Regret Certificate
06:00 · July 14, 2026 · arXiv cs.AI RSS
Language models turn a worded situation into a numeric plan, and the dominant pipelines (NL4Opt, OptiMUS, ORLM, OR-LLM-Agent) commit to a single objective and point-valued coefficients, then solve once. For decisions that allocate real budget, effort, or clinical attention, that confidence is the failure mode: every objectified number is an assumption, and a plan optimal only if the guesses are exactly right is fragile -- mimicry of computation. YUKTI changes the target of autoformulation. Its representation is a typed-proposition graph whose relationships carry shape priors, coefficient uncertainty, and provenance. YUKTI routes each stage to an exact, nonlinear, or evolutionary solver; couples stages by a distributional Pareto hand-off; and introduces Assumption-Robust Pareto Frontiers (ARPF), resampling assumptions (including structural epsilon-contamination) to score how often each action survives (rho). We prove a bound making rho an exact factor of decision regret, add auditable traceability, and synthesize a benchmark-faithful data foundation when none exists (SRJANA). We validate three ways: under controlled misspecification the robust compromise cuts mean and tail regret by over 90% versus a naive point plan; on a regulated commercial decision we optimize inside a lawful action space and price the downside in euros; and on a real public dataset of 41,188 decisions an out-of-sample backtest beats the logged status quo by 34% and a naive point rule by 4% while reducing the optimizer's curse. The solvers are standard; we claim no benchmark-SOTA win. A head-to-head shows an LLM given the correct numbers, and single-objective optimization, both incur about 47x the held-out regret of YUKTI -- an LLM is a formulator, not a solver. Under long-range causal coupling, the forward hand-off becomes unsound, locating where it must become a backward-induction causal policy.
Summary
YUKTI addresses a core limitation in current LLM-driven optimization pipelines, which typically extract a single objective and fixed coefficients from natural-language descriptions before emitting one solver-ready program. Such point-valued formulations treat every elicited number as exact, leaving decisions vulnerable when those assumptions deviate from reality. The framework instead produces a Typed Proposition intermediate representation in which each quantitative relationship carries a shape prior, a distribution over its coefficients, and a provenance tag indicating whether the value is given, assumed, or benchmark-derived.
A structure-aware router then inspects each stage numerically and dispatches it to an exact, nonlinear, or evolutionary multi-objective solver. Stages are coupled through distributional Pareto hand-offs that propagate uncertainty forward rather than collapsing it to a single compromise at each step. Assumption-Robust Pareto Frontiers (ARPF) resample the coefficient distributions, including controlled structural misspecification via ε-contamination, and compute for every candidate solution the probability ρ that it remains feasible and non-dominated. A proven regret bound establishes ρ as an exact scaling factor of pool regret, supplying a formal certificate of fragility.
Decision traceability is obtained through segment attribution and shadow-price reporting, which identify the constituent segments of a recommended action and the binding constraints. When no empirical dataset exists, a front-end module (SRJANA) synthesizes a benchmark-anchored context to fit the propositions. Validation on controlled misspecification tests, a regulated oncology brand pricing exercise, and an out-of-sample backtest over 41,188 real marketing decisions shows the robust compromise materially reducing both mean and tail regret relative to point-valued baselines while preserving auditability. The system is positioned as a stress-testing layer rather than a replacement for existing solvers, underscoring that language models are effective at formulation but not at solving under uncertainty.
Why it matters
High technical depth and novelty in robust autoformulation directly address EU-regulated decision systems; Dutch AI researchers can apply the ARPF mechanism and regret certificate to build auditable optimization layers for commercial or public-sector use.
