Dataset Information

Using simulation to accelerate autonomous experimentation: A case study using mechanics.

ABSTRACT: Autonomous experimentation (AE) accelerates research by combining automation and machine learning to perform experiments intelligently and rapidly in a sequential fashion. While AE systems are most needed to study properties that cannot be predicted analytically or computationally, even imperfect predictions can in principle be useful. Here, we investigate whether imperfect data from simulation can accelerate AE using a case study on the mechanics of additively manufactured structures. Initially, we study resilience, a property that is well-predicted by finite element analysis (FEA), and find that FEA can be used to build a Bayesian prior and experimental data can be integrated using discrepancy modeling to reduce the number of needed experiments ten-fold. Next, we study toughness, a property not well-predicted by FEA and find that FEA can still improve learning by transforming experimental data and guiding experiment selection. These results highlight multiple ways that simulation can improve AE through transfer learning.

SUBMITTER: Gongora AE

PROVIDER: S-EPMC8010472 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:Spatial capture-recapture (SCR) models have improved the ability to estimate densities of rare and elusive animals. However, SCR models have seldom been validated even as model formulations diversify and expand to incorporate new sampling methods and/or additional sources of information on model parameters. Information on the relationship between encounter probabilities, sources of additional information, and the reliability of density estimates, is rare but crucial to assessing reliability of SCR-based estimates. We used a simulation-based approach that incorporated prior empirical work to assess the accuracy and precision of density estimates from SCR models using spatially unstructured sampling. To assess the consequences of sparse data and potential sources of bias, we simulated data under six scenarios corresponding to three different levels of search effort and two levels of correlation between search effort and animal density. We then estimated density for each scenario using four models that included increasing amounts of information from harvested individuals and telemetry to evaluate the impact of additional sources of information. Model results were sensitive to the quantity of available information: density estimates based on low search effort were biased high and imprecise, whereas estimates based on high search effort were unbiased and precise. A correlation between search effort and animal density resulted in a positive bias in density estimates, though the bias decreased with increasingly informative datasets. Adding information from harvested individuals and telemetered individuals improved density estimates based on low and moderate effort but had negligible impact for datasets resulting from high effort. We demonstrated that density estimates from SCR models using spatially unstructured sampling are reliable when sufficient information is provided. Accurate density estimates can result if empirical-based simulations such as those presented here are used to develop study designs with appropriate amounts of effort and information sources.

Dataset Information

Using simulation to accelerate autonomous experimentation: A case study using mechanics.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets