Optimizing a reinforcement learning (RL) policy typically requires extensive interactions with a high-fidelity simulator of the environment, which are often costly or impractical. Offline RL addresses this problem by allowing training from pre-collected data, but its effectiveness is strongly constrained by the size and quality of the dataset. Hybrid offline-online RL leverages both offline data and interactions with a single simulator of the environment. In many real-world scenarios, however, multiple simulators with varying levels of fidelity and computational cost are available. In this work, we study multi-fidelity hybrid RL for policy optimization under a fixed cost budget. We introduce multi-fidelity hybrid RL via information gain maximization (MF-HRL-IGM), a hybrid offline-online RL algorithm that implements fidelity selection based on information gain maximization through a bootstrapping approach. Theoretical analysis establishes the no-regret property of MF-HRL-IGM, while empirical evaluations demonstrate its superior performance compared to existing benchmarks.

Determining the optimal data mixture for large language model training remains a challenging problem with an outsized impact on performance. In practice, language model developers continue to rely on heuristic exploration since no learning-based approach has emerged as a reliable solution. In this work, we propose to view the selection of training data mixtures as a black-box hyperparameter optimization problem, for which Bayesian Optimization is a well-established class of appropriate algorithms. Firstly, we cast data mixture learning as a sequential decision-making problem, in which we aim to find a suitable trade-off between the computational cost of training exploratory (proxy-) models and final mixture performance. Secondly, we systematically explore the properties of transferring mixtures learned at a small scale to larger-scale experiments, providing insights and highlighting opportunities for research at a modest scale. By proposing Multi-fidelity Bayesian Optimization as a suitable method in this common scenario, we introduce a natural framework to balance experiment cost with model fit, avoiding the risks of overfitting to smaller scales while minimizing the number of experiments at high cost. We present results for pre-training and instruction finetuning across models ranging from 1 million to 7 billion parameters, varying from simple architectures to state-of-the-art models and benchmarks spanning dozens of datasets. We demonstrate consistently strong results relative to a wide range of baselines, resulting inspeed-ups of over 500% in determining the best data mixture on our largest experiments. In addition, we broaden access to research by sharing ADMIRE IFT Runs, a dataset of 460 full training & evaluation runs worth over 13,000 GPU hours, greatly reducing the cost of conducting research in this area.

Bioprocesses are central to modern biotechnology, enabling sustainable production in pharmaceuticals, specialty chemicals, cosmetics, and food. However, developing high-performing processes is costly and complex, requiring iterative, multi-scale experimentation from microtiter plates to pilot reactors. Conventional Design of Experiments (DoE) approaches often struggle to address process scale-up and the joint optimization of reaction conditions and biocatalyst selection. We propose a multi-fidelity batch Bayesian optimization framework to accelerate bioprocess development and reduce experimental costs. The method integrates Gaussian Processes tailored for multi-fidelity modeling and mixed-variable optimization, guiding experiment selection across scales and biocatalysts. A custom simulation of a Chinese Hamster Ovary bioprocess, capturing non-linear and coupled scale-up dynamics, is used for benchmarking against multiple simulated industrial DoE baselines. Multiple case studies show how the proposed workflow can achieve a reduction in experimental costs and increased yield. This work provides a data-efficient strategy for bioprocess optimization and highlights future opportunities in transfer learning and uncertainty-aware design for sustainable biotechnology.

The simulation cost for cosmological simulation-based inference can be decreased by combining simulation sets of varying fidelity. We propose an approach to such multi-fidelity inference based on feature matching and knowledge distillation. Our method results in improved posterior quality, particularly for small simulation budgets and difficult inference problems.

Parameter tuning for vehicle controllers remains a costly and time-intensive challenge in automotive development. Traditional approaches rely on extensive real-world testing, making the process inefficient. We propose a multi-fidelity Bayesian optimization approach that efficiently learns optimal controller parameters by leveraging both low-fidelity simulation data and a very limited number of real-world experiments. Our approach significantly reduces the need for manual tuning and expensive field testing while maintaining the standard two-stage development workflow used in industry. The core contribution is the integration of an auto-regressive multi-fidelity Gaussian process model into Bayesian optimization, enabling knowledge transfer between different fidelity levels without requiring additional low-fidelity evaluations during real-world testing. We validate our approach through both simulation studies and realworld experiments. The results demonstrate that our method achieves high-quality controller performance with only very few real-world experiments, highlighting its potential as a practical and scalable solution for intelligent vehicle control tuning in industrial applications.

Multi-fidelity Reinforcement Learning (RL) frameworks significantly enhance the efficiency of engineering design by leveraging analysis models with varying levels of accuracy and computational costs. The prevailing methodologies, characterized by transfer learning, human-inspired strategies, control variate techniques, and adaptive sampling, predominantly depend on a structured hierarchy of models. However, this reliance on a model hierarchy overlooks the heterogeneous error distributions of models across the design space, extending beyond mere fidelity levels. This work proposes ALPHA (Adaptively Learned Policy with Heterogeneous Analyses), a novel multi-fidelity RL framework to efficiently learn a high-fidelity policy by adaptively leveraging an arbitrary set of non-hierarchical, heterogeneous, low-fidelity models alongside a high-fidelity model. Specifically, low-fidelity policies and their experience data are dynamically used for efficient targeted learning, guided by their alignment with the high-fidelity policy. The effectiveness of ALPHA is demonstrated in analytical test optimization and octocopter design problems, utilizing two low-fidelity models alongside a high-fidelity one. The results highlight ALPHA's adaptive capability to dynamically utilize models across time and design space, eliminating the need for scheduling models as required in a hierarchical framework. Furthermore, the adaptive agents find more direct paths to high-performance solutions, showing superior convergence behavior compared to hierarchical agents.

A critical bottleneck for scientific progress is the costly nature of computer simulations for complex systems. Surrogate models provide an appealing solution: such models are trained on simulator evaluations, then used to emulate and quantify uncertainty on the expensive simulator at unexplored inputs. In many applications, one often has available data on related systems. For example, in designing a new jet turbine, there may be existing studies on turbines with similar configurations. A key question is how information from such "source" systems can be transferred for effective surrogate training on the "target" system of interest. We thus propose a new LOcal transfer Learning Gaussian Process (LOL-GP) model, which leverages a carefully-designed Gaussian process to transfer such information for surrogate modeling. The key novelty of the LOL-GP is a latent regularization model, which identifies regions where transfer should be performed and regions where it should be avoided. This "local transfer" property is desirable in scientific systems: at certain parameters, such systems may behave similarly and thus transfer is beneficial; at other parameters, they may behave differently and thus transfer is detrimental. By accounting for local transfer, the LOL-GP can rectify a critical limitation of "negative transfer" in existing transfer learning models, where the transfer of information worsens predictive performance. We derive a Gibbs sampling algorithm for efficient posterior predictive sampling on the LOL-GP, for both the multi-source and multi-fidelity transfer settings. We then show, via a suite of numerical experiments and an application for jet turbine design, the improved surrogate performance of the LOL-GP over existing methods.

Current generative models for drug discovery primarily use molecular docking as an oracle to guide the generation of active compounds. However, such models are often not useful in practice because even compounds with high docking scores do not consistently show experimental activity. More accurate methods for activity prediction exist, such as molecular dynamics based binding free energy calculations, but they are too computationally expensive to use in a generative model. To address this challenge, we propose Multi-Fidelity Latent space Active Learning (MF-LAL), a generative modeling framework that integrates a set of oracles with varying cost-accuracy tradeoffs. Unlike previous approaches that separately learn the surrogate model and generative model, MF-LAL combines the generative and multi-fidelity surrogate models into a single framework, allowing for more accurate activity prediction and higher quality samples. We train MF-LAL with a novel active learning algorithm to further reduce computational cost. Our experiments on two disease-relevant proteins show that MF-LAL produces compounds with significantly better binding free energy scores than other single and multi-fidelity approaches.

Testing controllers in safety-critical systems is vital for ensuring their safety and preventing failures. In this paper, we address the falsification problem within learning-based closed-loop control systems through simulation. This problem involves the identification of counterexamples that violate system safety requirements and can be formulated as an optimization task based on these requirements. Using full-fidelity simulator data in this optimization problem can be computationally expensive. To improve efficiency, we propose a multi-fidelity Bayesian optimization falsification framework that harnesses simulators with varying levels of accuracy. Our proposed framework can transition between different simulators and establish meaningful relationships between them. Through multi-fidelity Bayesian optimization, we determine both the optimal system input likely to be a counterexample and the appropriate fidelity level for assessment. We evaluated our approach across various Gym environments, each featuring different levels of fidelity. Our experiments demonstrate that multi-fidelity Bayesian optimization is more computationally efficient than full-fidelity Bayesian optimization and other baseline methods in detecting counterexamples. A Python implementation of the algorithm is available at https://github.com/SAILRIT/MFBO_Falsification.