The Value of Mechanistic Priors in Sequential Decision Making

• URL: http://arxiv.org/abs/2605.10018v1
• Date: Mon, 11 May 2026 05:43:53 GMT
• Title: The Value of Mechanistic Priors in Sequential Decision Making
• Abstract summary: Hybrid mechanistic models, physical priors with learned residuals, promise to reduce the data required for good decisions, but have no computable criterion to test this.<n>We characterize the value of mechanistic priors in sequential decision-making within both and burn-in regimes.
• Score: 41.418304743101174
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Hybrid mechanistic models, physical priors with learned residuals, promise to reduce the data required for good decisions, but have no computable criterion to test this. We characterize the value of mechanistic priors in sequential decision-making within both asymptotic and burn-in regimes. To formalize this, we introduce the mechanistic information of a model -- the mutual information between the model's recommended policy $\hatπ$ and the true optimal policy $π^*$ -- quantified via an occupancy-weighted bias $B_μ$. In the asymptotic regime (large $N$), matched bounds reveal that Bayesian regret scales with the residual entropy $H_{\mathrm{mech}}$, delivering a theoretical sample complexity reduction of $H(μ)/H_{\mathrm{mech}}$ compared to an uninformed baseline. Furthermore, we provide a model certificate to determine empirical sample efficiency. Complementarily, in the clinically relevant burn-in regime (small $N$), we establish a lower bound on the penalty incurred by confidently wrong priors. We demonstrate both the asymptotic and burn-in bounds across 5-fluorouracil (5-FU) dosing simulations motivated by published FOLFOX pharmacokinetic data, where a hybrid prior yields large sample-efficiency gains in the burn-in regime. Finally, we contrast these grounded models with LLM priors, demonstrating that LLMs can suffer severe losses in mechanistic information, thereby motivating the exclusive use of physically-grounded priors for safety-critical applications.

Related papers

• Provably Reduced Sample Cost in Prior-Guided Hyperparameter Optimization [14.49729393385318]
  We provide the first distribution-dependent sample complexity bounds for multi-fidelity HPO with priors.<n>We show that informative priors concentrate probability mass on near-optimal arms, yield reductions in the number of required evaluations, whereas baseline performance is recovered with uninformative or misleading priors.
  arXiv  Detail & Related papers  (2026-06-03T13:32:17Z)
• $V_{0.5}$: Generalist Value Model as a Prior for Sparse RL Rollouts [81.48669089692189]
  Generalist Value Models (such as $V_0.5$) achieve pre-trained value estimation by explicitly encoding model capabilities in-context.<n>We propose $V_0.5$, which adaptively fuses the baseline predicted by such value model with the empirical mean derived from sparse rollouts.<n>$V_0.5$ significantly outperforms GRPO and DAPO, achieving faster convergence and over some 10% performance improvement.
  arXiv  Detail & Related papers  (2026-03-11T14:57:41Z)
• On the Asymptotic Mean Square Error Optimality of Diffusion Models [10.72484143420088]
  Diffusion models (DMs) as generative priors have recently shown great potential for denoising tasks.<n>This paper proposes a novel denoising strategy inspired by the structure of the MSE-optimal conditional mean (CME)<n>The resulting DM-based denoiser can be conveniently employed using a pre-trained DM, being particularly fast by truncating reverse diffusion steps.
  arXiv  Detail & Related papers  (2024-03-05T13:25:44Z)
• A Pseudo-Semantic Loss for Autoregressive Models with Logical Constraints [87.08677547257733]
  Neuro-symbolic AI bridges the gap between purely symbolic and neural approaches to learning. We show how to maximize the likelihood of a symbolic constraint w.r.t the neural network's output distribution. We also evaluate our approach on Sudoku and shortest-path prediction cast as autoregressive generation.
  arXiv  Detail & Related papers  (2023-12-06T20:58:07Z)
• Self-Supervised Dataset Distillation for Transfer Learning [77.4714995131992]
  We propose a novel problem of distilling an unlabeled dataset into a set of small synthetic samples for efficient self-supervised learning (SSL) We first prove that a gradient of synthetic samples with respect to a SSL objective in naive bilevel optimization is textitbiased due to randomness originating from data augmentations or masking. We empirically validate the effectiveness of our method on various applications involving transfer learning.
  arXiv  Detail & Related papers  (2023-10-10T10:48:52Z)
• Transfer Learning for Causal Effect Estimation [12.630663215983706]
  We present a Transfer Causal Learning framework to improve causal effect estimation accuracy in limited data. Our method is subsequently extended to real data and generates meaningful insights consistent with medical literature.
  arXiv  Detail & Related papers  (2023-05-16T03:13:55Z)
• Maximum Likelihood Learning of Unnormalized Models for Simulation-Based Inference [44.281860162298564]
  We introduce two synthetic likelihood methods for Simulation-Based Inference. We learn a conditional energy-based model (EBM) of the likelihood using synthetic data generated by the simulator. We demonstrate the properties of both methods on a range of synthetic datasets, and apply them to a model of the neuroscience network in the crab.
  arXiv  Detail & Related papers  (2022-10-26T14:38:24Z)
• Inverting brain grey matter models with likelihood-free inference: a tool for trustable cytoarchitecture measurements [62.997667081978825]
  characterisation of the brain grey matter cytoarchitecture with quantitative sensitivity to soma density and volume remains an unsolved challenge in dMRI. We propose a new forward model, specifically a new system of equations, requiring a few relatively sparse b-shells. We then apply modern tools from Bayesian analysis known as likelihood-free inference (LFI) to invert our proposed model.
  arXiv  Detail & Related papers  (2021-11-15T09:08:27Z)
• Machine Learning Product State Distributions from Initial Reactant States for a Reactive Atom-Diatom Collision System [2.678461526933908]
  A machine learned (ML) model for predicting product state distributions from specific initial states is presented. The prediction accuracy as quantified by the root-mean-squared difference is high for the test set and off-grid state specific initial conditions. The STD model can be well-suited for simulating nonequilibrium high-speed flows.
  arXiv  Detail & Related papers  (2021-11-05T15:36:27Z)
• Variational Inference with NoFAS: Normalizing Flow with Adaptive Surrogate for Computationally Expensive Models [7.217783736464403]
  Use of sampling-based approaches such as Markov chain Monte Carlo may become intractable when each likelihood evaluation is computationally expensive. New approaches combining variational inference with normalizing flow are characterized by a computational cost that grows only linearly with the dimensionality of the latent variable space. We propose Normalizing Flow with Adaptive Surrogate (NoFAS), an optimization strategy that alternatively updates the normalizing flow parameters and the weights of a neural network surrogate model.
  arXiv  Detail & Related papers  (2021-08-28T14:31:45Z)
• Amortized Conditional Normalized Maximum Likelihood: Reliable Out of Distribution Uncertainty Estimation [99.92568326314667]
  We propose the amortized conditional normalized maximum likelihood (ACNML) method as a scalable general-purpose approach for uncertainty estimation. Our algorithm builds on the conditional normalized maximum likelihood (CNML) coding scheme, which has minimax optimal properties according to the minimum description length principle. We demonstrate that ACNML compares favorably to a number of prior techniques for uncertainty estimation in terms of calibration on out-of-distribution inputs.
  arXiv  Detail & Related papers  (2020-11-05T08:04:34Z)
• On a computationally-scalable sparse formulation of the multidimensional and non-stationary maximum entropy principle [0.0]
  We derive a non-stationary formulation of the MaxEnt-principle and show that its solution can be approximated through a numerical maximisation of the sparse constrained optimization problem with regularization. We show that all of the considered seven major financial benchmark time series are better described by conditionally memoryless MaxEnt-models. This analysis also reveals a sparse network of statistically-significant temporal relations for the positive and negative latent variance changes among different markets.
  arXiv  Detail & Related papers  (2020-05-07T05:22:46Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.