Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “prediction uncertainty”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

An overview of wind-energy-production prediction bias, losses, and uncertainties

Abstract. The financing of a wind farm directly relates to the preconstruction energy yield assessments which estimate the annual energy production for the farm. The accuracy and the precision of the preconstruction energy estimates can dictate the profitability of the wind project. Historically, the wind industry tended to overpredict the annual energy production of wind farms. Experts have been dedicated to eliminating such prediction errors in the past decade, and recently the reported average energy prediction bias is declining. Herein, we present a literature review of the energy yield assessment errors across the global wind energy industry. We identify a long-term trend of reduction in the overprediction bias, whereas the uncertainty associated with the prediction error is prominent. We also summarize the recent advancements of the wind resource assessment process that justify the bias reduction, including improvements in modeling and measurement techniques. Additionally, because the energy losses and uncertainties substantially influence the prediction error, we document and examine the estimated and observed loss and uncertainty values from the literature, according to the proposed framework in the International Electrotechnical Commission 61400-15 wind resource assessment standard. From our findings, we highlight opportunities for the industry to move forward, such as the validation and reduction of prediction uncertainty and the prevention of energy losses caused by wake effect and environmental events. Overall, this study provides a summary of how the wind energy industry has been quantifying and reducing prediction errors, energy losses, and production uncertainties. Finally, for this work to be as reproducible as possible, we include all of the data used in the analysis in appendices to the article.

17 WIND ENERGY↗
DOE OSTI2021

Development of a framework for sequential Bayesian design of experiments: Application to a pilot-scale solvent-based CO 2 capture process

In this paper, a methodology is developed for sequential design of experiments (SDoE) for process systems and applied to a solvent-based CO 2 capture system. In this approach, the prior knowledge of the system is used to prioritize process data collection at specific operating conditions. These data are then incorporated into a Bayesian inference methodology for updating a stochastic model by refining estimations of its underlying parameters, and the updated model is then used to generate the next set of test runs. Thus, the new knowledge obtained from the data is used to guide subsequent iterations of the experimental runs, ensuring that the overall data collection is maximally informative given that most experimental campaigns, especially at pilot or higher-scale plants, are costly, time-consuming, and resource-limited. The test run objective for this work was to minimize the maximum model prediction uncertainty for key output variables, but the methodology is generic and can be readily applied to other test run objectives. This methodology is applied to an aqueous monoethanolamine (MEA) pilot plant campaign at the National Carbon Capture Center (NCCC) in Wilsonville, Alabama, USA. The SDoE framework was utilized for two iterations, while collecting 18 sets of data representing different process conditions, and this resulted in an overall average reduction in uncertainty of approximately 50% in the prediction of CO 2 capture percentage. Moreover, 11 additional data sets were obtained with variation of absorber packing height for further model validation. This work shows the capability of the SDoE framework to maximize learning given limited resources, allowing for the reduction of model uncertainty, which is of great importance for many applications including reduction of technical risk associated with scale-up and economic analysis.

20 FOSSIL-FUELED POWER PLANTS↗
DOE OSTI2022

Explainable Bayesian Neural Network for Probabilistic Transient Stability Analysis Considering Wind Energy

While several data-driven models have been developed for transient stability assessment, how to consider the uncertainties from load and renewable generations and provide interpretation of data-driven assessment results are still open. This paper proposes an explainable Bayesian Neural Network (BNN) for probabilistic transient stability assessment (TSA). By extracting the uncertainties from loads and wind farms, the BNN model can make a reliable prediction and quantify the prediction uncertainties. We also develop the Gradient Shap algorithm to make the global and local explanations for the probabilistic TSA model, a significant advantage over existing black-box data-driven methods. Numerical results on the modified IEEE 39-bus system show that the proposed method outperforms the existing methods in terms of prediction accuracy and uncertainty quantification capabilities. The explainability of the proposed method allows system operators to design preventive controls for enhancing system stability.

Bayesian Neural Network↗
DOE OSTI2025

LTAU-FF: Loss Trajectory Analysis for Uncertainty in atomistic Force Fields

Model ensembles are effective tools for estimating prediction uncertainty in deep learning atomistic force fields. However, their widespread adoption is hindered by high computational costs and overconfident error estimates. In this work, we address these challenges by leveraging distributions of per-sample errors obtained during training and employing a distance-based similarity search in the model latent space. Our method, which we call LTAU (Loss Trajectory Analysis for Uncertainty), efficiently estimates the full probability distribution function of errors for any test point using the logged training errors, achieving speeds that are 2–3 orders of magnitudes faster than typical ensemble methods and allowing it to be used for tasks where training or evaluating multiple models would be infeasible. We apply LTAU towards estimating parametric uncertainty in atomistic force fields (LTAU-FF), demonstrating that it produces well-calibrated confidence intervals and predicts errors that correlate strongly with the true errors for data near the training domain. Furthermore, we show that the errors predicted by LTAU-FF can be used in practical applications for detecting out-of-domain data, tuning model performance, and predicting failure during simulations. We believe that LTAU will be a valuable tool for uncertainty quantification in atomistic force fields and is a promising method that should be further explored in other domains of machine learning.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Selection of high-redshift Lyman-Break Galaxies from broadband and wide photometric surveys

Here, in this paper, we investigate the possibility of selecting high-redshift Lyman-Break Galaxies (LBG) using current and future broadband wide photometric surveys, such as the Ultraviolet Near Infrared Optical Northern Survey (UNIONS) or the Vera C. Rubin Legacy Survey of Space and Time (LSST), using a Random Forest algorithm. This work is conducted in the context of future large-scale structure spectroscopic surveys like DESI-II, the next phase of the Dark Energy Spectroscopic Instrument (DESI), which will start around 2029. We use deep imaging data from the Hyper Suprime Camera (HSC) and the Canada-France-Hawaii Telescope Large Area U-band Deep Survey (CLAUDS) on the COSMOS and XMM-LSS fields. To predict the selection performance of LBGs with image quality similar to UNIONS, we degrade the u,g,r,i and z bands to UNIONS depth. The Random Forest algorithm is trained with the u,g,r,i and z bands to classify LBGs in the 2.5 < z < 3.5 range. We find that fixing a target density budget of 1,100 deg -2 , the Random Forest approach gives a density of z > 2 targets of 873 deg -2 , and a density of 493 deg -2 of confirmed LBGs after spectroscopic confirmation with DESI. This UNIONS-like selection was tested in a dedicated spectroscopic observation campaign of 1,000 targets with DESI on the COSMOS field, providing a safe spectroscopic sample with a mean redshift of 3. This sample is used to derive forecasts for DESI-II, assuming a sky coverage of 5,000 deg 2 . We predict uncertainties on Alcock-Paczynski parameters α ⊥ and α ∥ to be 0.7% and 1% for 2.6 < z < 3.2, resulting in a potential 2% measurement of the dark energy fraction at high redshift. Additionally, we estimate the uncertainty in local non-Gaussianity and predict σ fNL ≈ 7, which would be comparable to the current best precision achieved by Planck. The latter forecast suggests that achieving the precision required to place stringent constraints on inflationary models (σ fNL ≈ 1) using spectroscopic galaxy surveys necessitates the development of a next-generation (Stage V) spectroscopic survey.

cosmological parameters from LSS↗
DOE OSTI2025

Selection of high-redshift Lyman-Break Galaxies from broadband and wide photometric surveys

In this paper, we investigate the possibility of selecting high-redshift Lyman-Break Galaxies (LBG) using current and future broadband wide photometric surveys, such as the Ultraviolet Near Infrared Optical Northern Survey (UNIONS) or the Vera C. Rubin Legacy Survey of Space and Time (LSST), using a Random Forest algorithm. This work is conducted in the context of future large-scale structure spectroscopic surveys like DESI-II, the next phase of the Dark Energy Spectroscopic Instrument (DESI), which will start around 2029.We use deep imaging data from the Hyper Suprime Camera (HSC) and the Canada-France-Hawaii Telescope Large Area U-band Deep Survey (CLAUDS) on the COSMOS and XMM-LSS fields. To predict the selection performance of LBGs with image quality similar to UNIONS, we degrade the u,g,r,i and z bands to UNIONS depth.The Random Forest algorithm is trained with the u,g,r,i and z bands to classify LBGs in the 2.5 < z < 3.5 range.We find that fixing a target density budget of 1,100 deg$^{-2}$, the Random Forest approach gives a density of z > 2 targets of 873 deg$^{-2}$, and a density of 493 deg$^{-2}$ of confirmed LBGs after spectroscopic confirmation with DESI. This UNIONS-like selection was tested in a dedicated spectroscopic observation campaign of 1,000 targets with DESI on the COSMOS field, providing a safe spectroscopic sample with a mean redshift of 3. This sample is used to derive forecasts for DESI-II, assuming a sky coverage of 5,000 deg$^{2}$. We predict uncertainties on Alcock-Paczynski parameters α$_{⊥}$ and α$_{∥}$ to be 0.7% and 1% for 2.6 < z < 3.2, resulting in a potential 2% measurement of the dark energy fraction at high redshift. Additionally, we estimate the uncertainty in local non-Gaussianity and predict σ$_{fNL}$ ≈ 7, which would be comparable to the current best precision achieved by Planck. The latter forecast suggests that achieving the precision required to place stringent constraints on inflationary models (σ$_{fNL}$ ≈ 1) using spectroscopic galaxy surveys necessitates the development of a next-generation (Stage V) spectroscopic survey.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

GPLaSDI: Gaussian Process Latent Space Dynamics Identification

GPLaSDI is a software that introduces GPLaSDI, a novel LaSDI-based framework that relies on Gaussian process (GP) for latent space ODE interpolations. Using GPs offers two significant advantages. First, it enables the quantification of uncertainty over the reduced order model (ROM) predictions. Second, leveraging this prediction uncertainty allows for efficient adaptive training through a greedy selection of additional training data points. This approach does not require prior knowledge of the underlying PDEs. Consequently, GPLaSDI is inherently non-intrusive and can be applied to problems without a known PDE or its residual. We demonstrate the effectiveness of our approach on the Burgers equation, Vlasov equation for plasma physics, and a rising thermal bubble problem. Our proposed method achieves between 200 and 100,000 times speed-up, with up to 7% relative error.

Choi, Youngsoo↗
DOE OSTI2019

Deep Direct-Use Feasibility Study Numerical Modeling and Uncertainty Analysis using iTOUGH2 for West Virginia University

To reduce the geothermal exploration risk, a feasibility study is performed for a deep direct-use system proposed at the West Virginia University (WVU) Morgantown campus. This study applies numerical simulations to investigate reservoir impedance and thermal production. Because of the great depth of the geothermal reservoir, few data are available to characterize reservoir features and properties. As a result, the study focuses on the following three aspects: 1. model choice for predicting reservoir impedance and thermal breakthrough: after investigating three potential models (one single permeability model and two dual permeability models) for flow through fractured rock, it is decided to use single permeability model for further analysis; 2. well placement (horizontal vs. vertical) options: horizontal well placement seems to be more robust to heterogeneity and the impedance is more acceptable; 3. Prediction uncertainty: the most influential parameters are identified using a First-Order-Second-Moment uncertainty propagation analysis, and the uncertain range of the model predictions is estimated by performing a Monte Carlo simulation. Heterogeneity has a large impact on the perdition, therefore, is considered in the predictive model and uncertainty analysis. The numerical model results and uncertainty analysis are used for economic analysis. The dataset submitted here support the described study. Manuscript is submitted to Geothermics, will be linked once paper is accepted.

15 GEOTHERMAL ENERGY↗
DOE OSTI2023

Spatial Bayesian models project shifts in suitable habitat for Pacific Northwest tree species under climate change

Abstract We developed spatial Bayesian hierarchical models to assess potential climate change impacts on suitable habitat for five important tree species in the Pacific northwestern United States (California, Oregon, and Washington). Individual‐species models were fit with presence–absence data from forest inventory field plots and spatial relationships were specified through a conditional autoregressive model. This modeling approach allowed us to visualize uncertainty in response curves, map current and future prediction uncertainty, and provide interval estimates for change. Upward elevational or northward latitudinal shifts in climatically suitable habitat were projected for all species. Climate change impacts were the most damaging for noble fir ( Abies procera ), for which 79%–100% of the current range was projected to become climatically unsuitable by the 2080s. Although coastal Douglas‐fir ( Pseudotsuga menziesii var. menziesii ) has been projected by others to gain habitat in Canada, within our study area we projected a net loss of climatically suitable habitat (ca. 8000–31,400 km 2 ) under three of four future climate scenarios. A net loss in habitat was also projected for Oregon white oak ( Quercus garryana ) under three of four scenarios, with 40%–60% of the current range becoming unsuitable. Although there was no net loss of habitat for forest land blue oak under any scenario, other factors like competition may inhibit blue oak ( Quercus douglasii ) and white oak from occupying areas projected to increase in climatic suitability. Additionally, between 13% and 32% of blue oak's current range was projected to become unsuitable; some of these areas aligned with dieback following the 2012–2015 California drought, which our data set predates. Unlike the other four species, we projected a 17%–25% increase in climatically suitable habitat for California black oak ( Quercus kelloggii ), although 1%–20% of the current range was still projected to become unsuitable. Our findings indicate that, although some species will face more pressure in tracking climatically suitable habitat than others, climate change will impact the location of suitable habitat for many species.

Kralicek, Karin↗
DOE OSTI2025

MOOSE ProbML: Parallelizable Probabilistic Machine Learning and Uncertainty Quantification Capabilities

The Multiphysics Object Oriented Simulation Environment (MOOSE) is a widely used open- source finite element software for performing multiphysics multiscale simulations in a massively parallel fashion. Recently, the computational team at Idaho National Laboratory (INL) has implemented Probabilistic Machine Learning (ProbML) capabilities in MOOSE—in a parallelized fashion—and enable active learning with large-scale computational models for tasks such as surrogate model development, scale bridging, forward/inverse uncertainty quantification (UQ), Bayesian optimization, etc. This presentation summarizes these developments in MOOSE along with demonstrations on several real applications relevant to nuclear energy. At the fundamental level, samplers like Monte Carlo/Latin Hypercube, variance reduction, parallelized Markov Chain Monte Carlo (MCMC) support uncertainty propagation in both forward and inverse settings. These samplers can be integrated with the Gaussian processes (GP) suite in MOOSE, which offer several variants like scalar GPs, multi-output GPs, and deep GPs, to enable active learning. These GPs can be tuned using gradient-based optimization methods like Adam and its variants or gradient-free methods like the elliptical slice sampler (a variant of MCMC adept under Gaussian settings) for more complex covariance kernels or likelihoods whose gradient computations can be cumbersome. A variety of batch acquisition functions permit parallelized evaluation of the computational model and support different learning objectives with high efficiency like Bayesian inference, global surrogate development, optimization, etc. Furthermore, libtorch integration supports training, evaluation, and re-training of neural networks and other complex machine learning models in active learning settings. The impacts of these developments are shown on several real applications: (1) nuclear fuel inverse UQ and model inadequacy assessment using the Kennedy O’Hagan framework; (2) uncertainty aware surrogate modeling for additive manufacturing to predict field quantities; (3) nuclear reactor rare events analysis; and (4) complex fluid flow prediction using a global surrogate with quantified prediction uncertainty. Finally, the outlook of MOOSE ProbML is discussed for both outer-loop and inner-loop computations in the broad view to accelerate fuels and materials qualification, address gaps in knowledge and data, and assess new reactor/fuel systems.

11 - NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2024

Sensitivity Analysis of Irradiated Fueled Experiments using the MOOSE Framework [Slides]

Modeling and simulation (M&S) methods are able to predict uncertainties in experimental parameters (e.g., power and fission density) during irradiation. A shortfall exists in predicting how sensitive some of the parameters will behave during the experimental process. Sensitivity and Uncertainty Quantification (SUQ) is critical in support of qualification and licensing reactor fuels. The application of a method to quantify the uncertainty in these experiments is critical to the prediction of their performance. In this work, we propose the use of a polynomial chaos expansion (PCE) method to quantify the sensitive parameters in these simulations and, in an extension, their experimental surrogates. We propose to perform M&S using PCE uncertainty quantification on a previously irradiated fueled experiment in order to provide a validation case for Griffin and expand its use as a verification and validation (V&V) tool for experiments with a neutronics component. Griffin is an advanced, deterministic neutronics analysis code built using the MOOSE (multiphysics object-oriented simulation environment) framework which can provide state-of-the-art neutronic analysis on M&S of experiments. We will use the stochastic tools module (STM) in MOOSE to provide PCE uncertainty quantification on the proposed experimental setup. Idaho National Laboratory (INL) does not yet have an in-house developed code with V&V approval for experiments performed on-site; this work would provide a necessary addition of support for experiments performed at INL. The Nuclear Regulatory Commission (NRC) has explicitly requested uncertainties in calculated values such as fuel power and burnup, and the development of this capability would benefit the relationship between INL and the NRC.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2023

Correlated Trajectory Uncertainty for Adaptive Sequential Decision Making

One of the great challenges with decision making tasks on real world systems is the fact that data is sparse and acquiring additional data is expensive. In these cases, it is often crucial to make a model of the environment to assist in making decisions. At the same time, limited data means that learned models are erroneous, making it just as important to equip the model with good predictive uncertainties. In the context of learning sequential decision making policies, these uncertainties can prove useful for informing which data to collect for the greatest improvement in policy performance \citep{mehta2021experimental, mehta2022exploration} or informing the policy about unsure regions of state and action space to avoid during test time \citep{yu2020mopo}. Additionally, assuming that realistic samples of the environment can be drawn, an adaptable policy can be trained that attempts to make optimal decisions for any given possible instance of the environment \citep{ghosh2022offline, chen2021offline}. In this work, we examine the so-called ``probabilistic neural network'' (PNN) model that is ubiquitous in model-based reinforcement learning (MBRL) works. We argue that while PNN models may have good marginal uncertainties, they form a distribution of non-smooth transition functions. Not only are these samples unrealistic and may hamper adaptability, but we also assert that this leads to poor uncertainty estimates when predicting multiple step trajectory estimates. To address this issue, we propose a simple sampling method that can be implemented on top of pre-existing models.We evaluate our sampling technique on a number of environments, including a realistic nuclear fusion task, and find that, not only do smooth transition function samples produce more calibrated uncertainties, but they also lead to better downstream performance for an adaptive policy.

Offline Reinforcement Learning↗
DOE OSTI2024

Distance preserving machine learning for uncertainty aware accelerator capacitance predictions

Abstract Accurate uncertainty estimations are essential for producing reliable machine learning models, especially in safety-critical applications such as accelerator systems. Gaussian process models are generally regarded as the gold standard for this task; however, they can struggle with large, high-dimensional datasets. Combining deep neural networks with Gaussian process approximation techniques has shown promising results, but dimensionality reduction through standard deep neural network layers is not guaranteed to maintain the distance information necessary for Gaussian process models. We build on previous work by comparing the use of the singular value decomposition against a spectral-normalized dense layer as a feature extractor for a deep neural Gaussian process approximation model and apply it to a capacitance prediction problem for the High Voltage Converter Modulators in the Oak Ridge Spallation Neutron Source. Our model shows improved distance preservation and predicts in-distribution capacitance values with less than 1% error.

43 PARTICLE ACCELERATORS↗
DOE OSTI2026

Low Activity Waste Glass Optimization with Property Models from Machine Learning, Part 2: Experimental Validation and Active Learning

The United States Department of Energy is responsible for managing legacy nuclear waste stored in underground tanks at the Hanford Site. To treat the waste, it is planned as the current baseline to separately vitrify low-activity waste (LAW) and high-level waste fractions. Previously, machine learning (ML) based glass property models (e.g., chemical durability, viscosity, electrical conductivity and SO3 solubility) were developed with prediction uncertainties. A waste glass optimization approach was then established to enable the capability of using these ML models in LAW glass formulation. In this study, the previous ML models were first experimentally validated, and the results were incorporated back into the database to update the ML models. The updated models and formulations showed increased waste loading while reducing the failure rate, demonstrating improved predictive accuracy, reduced uncertainties, and the effectiveness of active learning in guiding high-dimensional, nonlinear LAW glass design. This represents the first experimental validation of ML based LAW glass formulation, with practical benefits such as higher waste loading, shorter mission duration, and lower operational risk.

Lu, Xiaonan (ORCID:0000000179708148)↗
DOE OSTI2020

Multiclass classification experiments

We seek a multiclass classifier that can satisfy the following requirements. 1. Predict among at least three classes. 2. Handle a moderately large number of correlated predictors. 3. Handle mixed categorical and continuous predictors. 4. Handle missing values (in some way). 5. Be trained with about n = 1,000 responses, and 6. Quantify prediction uncertainty. Based on 5. above, it is assumed that accuracy of prediction is paramount (relative to, say, runtime). All methods considered in this document take on the order of minutes (most take seconds) on a 2.8 GHz Quad-Core Intel Core i7 processor with 16GB RAM. Secondly, quantification of uncertainty is assumed to be a secondary task, since, with this small data size, resampling methods (i.e. bootstrapping) are possible. Further studies should quantify the feasibility and accuracy of resampling methods, and/or other methods, for quantifying uncertainty. R was used for all software.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

An investigation on machine learning predictive accuracy improvement and uncertainty reduction using VAE-based data augmentation

The confluence of ultrafast computers with large memory, rapid progress in Machine Learning (ML) algorithms, and the availability of large datasets place multiple engineering fields at the threshold of dramatic progress. However, a unique challenge in nuclear engineering is data scarcity because experimentation on nuclear systems is usually more expensive and time-consuming than most other disciplines. One potential way to resolve the data scarcity issue is deep generative learning, which uses certain ML models to learn the underlying distribution of existing data and generate synthetic samples that resemble the real data. In this way, one can significantly expand the dataset to train more accurate predictive ML models. In this study, our objective is to evaluate the effectiveness of data augmentation using variational autoencoder (VAE)-based deep generative models. We investigated whether the data augmentation leads to improved accuracy in the predictions of a deep neural network (DNN) model trained using the augmented data. Additionally, the DNN prediction uncertainties are quantified using Bayesian Neural Networks (BNN) and conformal prediction (CP) to assess the impact on predictive uncertainty reduction. To test the proposed methodology, we used TRACE simulations of steady-state void fraction data based on the NUPEC Boiling Water Reactor Full-size Fine-mesh Bundle Test (BFBT) benchmark. Here, we found that augmenting the training dataset using VAEs has improved the DNN model’s predictive accuracy, improved the prediction confidence intervals, and reduced the prediction uncertainties.

Bayesian neural network↗
DOE OSTI2025

Parameter uncertainties for imperfect surrogate models in the low-noise regime

Abstract Bayesian regression determines model parameters by minimizing the expected loss, an upper bound to the true generalization error. However, this loss ignores model form error, or misspecification, meaning parameter uncertainties are significantly underestimated and vanish in the large data limit. As misspecification is the main source of uncertainty for surrogate models of low-noise calculations, such as those arising in atomistic simulation, predictive uncertainties are systematically underestimated. We analyze the true generalization error of misspecified, near-deterministic surrogate models, a regime of broad relevance in science and engineering. We show that posterior parameter distributions must cover every training point to avoid a divergence in the generalization error and design a compatible ansatz which incurs minimal overhead for linear models. The approach is demonstrated on model problems before application to thousand-dimensional datasets in atomistic machine learning. Our efficient misspecification-aware scheme gives accurate prediction and bounding of test errors in terms of parameter uncertainties, allowing this important source of uncertainty to be incorporated in multi-scale computational workflows.

Swinburne, Thomas D. (ORCID:0000000232554257)↗
DOE OSTI2023

Low-energy Electron-track Imaging for a Liquid Argon Time-projection-chamber Telescope Concept Using Probabilistic Deep Learning

The GammaTPC is an MeV-scale single-phase liquid argon time-projection-chamber gamma-ray telescope concept with a novel dual-scale pixel-based charge-readout system. It promises to enable a significant improvement in sensitivity to MeV-scale gamma rays over previous telescopes. The novel pixel-based charge readout allows for imaging of the tracks of electrons scattered by Compton interactions of incident gamma rays. The two primary contributors to the accuracy of a Compton telescope in reconstructing an incident gamma-ray’s original direction are its energy and position resolution. In this work, we focus on using deep learning to optimize the reconstruction of the initial position and direction of electrons scattered in Compton interactions, including using probabilistic models to estimate predictive uncertainty. We show that the deep-learning models are able to predict locations of Compton scatters of MeV-scale gamma rays from simulated 500 μm pixel-based data to better than 1 mm rms error and are sensitive to the initial direction of the scattered electron. We compare and contrast different deep-learning uncertainty estimation algorithms for reconstruction applications. Additionally, we show that event-by-event estimates of the uncertainty of the locations of the Compton scatters can be used to select those events that were reconstructed most accurately, leading to improvement in locating the origin of gamma-ray sources on the sky.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗