Search NASA⌕ Search

SEARCH · Search NASA

Results for “Domain knowledge”

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 181 records · Page 10

Theory-Guided Machine Learning in Materials Science

Materials scientists are increasingly adopting the use of machine learning tools to discover hidden trends in data and make predictions. Applying concepts from data science without foreknowledge of their limitations and the unique qualities of materials data, however, could lead to errant conclusions. The differences that exist between various kinds of experimental and calculated data require careful choices of data processing and machine learning methods. Here, we outline potential pitfalls involved in using machine learning without robust protocols. We address some problems of overfitting to training data using decision trees as an example, rational descriptor selection in the field of perovskites, and preserving physical interpretability in the application of dimensionality reducing techniques such as principal component analysis. We show how proceeding without the guidance of domain knowledge can lead to both quantitatively and qualitatively incorrect predictive models.

36 MATERIALS SCIENCE↗

A Review of Physics-Informed Machine Learning in Fluid Mechanics

Physics-informed machine-learning (PIML) enables the integration of domain knowledge with machine learning (ML) algorithms, which results in higher data efficiency and more stable predictions. This provides opportunities for augmenting—and even replacing—high-fidelity numerical simulations of complex turbulent flows, which are often expensive due to the requirement of high temporal and spatial resolution. In this review, we (i) provide an introduction and historical perspective of ML methods, in particular neural networks (NN), (ii) examine existing PIML applications to fluid mechanics problems, especially in complex high Reynolds number flows, (iii) demonstrate the utility of PIML techniques through a case study, and (iv) discuss the challenges and opportunities of developing PIML for fluid mechanics.

42 ENGINEERING↗

Application of Systems Engineering Principles and Techniques in Biological Big Data Analytics: A Review

In the past few decades, we have witnessed tremendous advancements in biology, life sciences and healthcare. These advancements are due in no small part to the big data made available by various high-throughput technologies, the ever-advancing computing power, and the algorithmic advancements in machine learning. Specifically, big data analytics such as statistical and machine learning has become an essential tool in these rapidly developing fields. As a result, the subject has drawn increased attention and many review papers have been published in just the past few years on the subject. Different from all existing reviews, this work focuses on the application of systems, engineering principles and techniques in addressing some of the common challenges in big data analytics for biological, biomedical and healthcare applications. Specifically, this review focuses on the following three key areas in biological big data analytics where systems engineering principles and techniques have been playing important roles: the principle of parsimony in addressing overfitting, the dynamic analysis of biological data, and the role of domain knowledge in biological data analytics.

dynamic analysis↗

Machine Reading at Scale: A Search Engine for Scientific and Academic Research

The Internet, much like our universe, is ever-expanding. Information, in the most varied formats, is continuously added to the point of information overload. Consequently, the ability to navigate this ocean of data is crucial in our day-to-day lives, with familiar tools such as search engines carving a path through this unknown. In the research world, articles on a myriad of topics with distinct complexity levels are published daily, requiring specialized tools to facilitate the access and assessment of the information within. Recent endeavors in artificial intelligence, and in natural language processing in particular, can be seen as potential solutions for breaking information overload and provide enhanced search mechanisms by means of advanced algorithms. As the advent of transformer-based language models contributed to a more comprehensive analysis of both text-encoded intents and true document semantic meaning, there is simultaneously a need for additional computational resources. Information retrieval methods can act as low-complexity, yet reliable, filters to feed heavier algorithms, thus reducing computational requirements substantially. In this work, a new search engine is proposed, addressing machine reading at scale in the context of scientific and academic research. It combines state-of-the-art algorithms for information retrieval and reading comprehension tasks to extract meaningful answers from a corpus of scientific documents. The solution is then tested on two current and relevant topics, cybersecurity and energy, proving that the system is able to perform under distinct knowledge domains while achieving competent performance.

Sousa, Norberto (ORCID:0000000329194817)↗

Gearbox bearing crack growth prognostics and uncertainty quantification with physics-informed machine learning

This paper introduces the extreme theory of functional connections (X-TFC), a physics-informed machine learning algorithm, and tailors it to estimate the remaining useful life (RUL) of wind turbine gearbox bearings experiencing fatigue crack growth. Unlike purely data-driven methods, X-TFC embeds a physics model, based on Head's theory in this work, into its training objective. The core of X-TFC is a random-projection single-layer neural network trained via an extreme learning machine, which requires only limited damage progression data and solves for output weights with a least-squares optimization algorithm. A composite loss function balances the network's fit to observed degradation data against the residuals of the governing crack growth differential equation, ensuring the learned damage trajectory remains physically plausible. When applied to a vibration-based health-index (HI) dataset measured during the growth of a crack on the inner ring of a high-speed bearing in a wind turbine gearbox (Bechhoefer and Dubé, 2020), X-TFC achieves near-zero prediction bias. Even when trained on only the first 10 %–20 % of the damage progression data, with sufficient physics weighting its predictions remain monotonic and smooth, delivering high prognosability and trendability. To quantify the epistemic uncertainty, we employ a Monte Carlo ensemble of independently initialized X-TFC models trained on noise-perturbed data, which yields confidence intervals around each RUL estimate and captures both model-parameter and epistemic uncertainty. In addition to a vibration-based HI, we demonstrate that the proposed framework can be directly applied to a supervisory control and data acquisition (SCADA) data-based HI (Eftekhari Milani et al., 2026) measured during similar wind turbine gearbox bearing crack faults, preserving its accuracy and interpretability. This extension shows the versatility of our approach, which is applicable to bearings of multiple gearbox manufacturers, models, and ratings using only SCADA data. By integrating domain knowledge with machine learning, X-TFC offers a rapid, reliable tool for crack prognostics. Its adaptability to other bearing failure modes, such as pitch bearing ring cracks, positions X-TFC as a powerful enabler of data-driven, physics-informed asset management in the wind energy sector and beyond.

17 WIND ENERGY↗

Decreasing wind speed extrapolation error via domain-specific feature extraction and selection

Abstract. Model uncertainty is a significant challenge in the wind energy industry and can lead to mischaracterization of millions of dollars' worth of wind resources. Machine learning methods, notably deep artificial neural networks (ANNs), are capable of modeling turbulent and chaotic systems and offer a promising tool to produce high-accuracy wind speed forecasts and extrapolations. This paper uses data collected by profiling Doppler lidars over three field campaigns to investigate the efficacy of using ANNs for wind speed vertical extrapolation in a variety of terrains, and it quantifies the role of domain knowledge in ANN extrapolation accuracy. A series of 11 meteorological parameters (features) are used as ANN inputs, and the resulting output accuracy is compared with that of both standard log-law and power-law extrapolations. It is found that extracted nondimensional inputs, namely turbulence intensity, current wind speed, and previous wind speed, are the features that most reliably improve the ANN's accuracy, providing up to a 65 % and 52 % increase in extrapolation accuracy over log-law and power-law predictions, respectively. The volume of input data is also deemed important for achieving robust results. One test case is analyzed in depth using dimensional and nondimensional features, showing that the feature nondimensionalization drastically improves network accuracy and robustness for sparsely sampled atmospheric cases.

17 WIND ENERGY↗

Development of Scanning Near-Field Ultrasound Holography as an Add-On Module for AFM systems

The project aimed to develop a standalone module for Atomic Force Microscope systems that the biological and environmental research community could use to image and quantify to a greater extent the interactions between nanoparticles and cells. If successful, it would enable the thousands of biological and environmental researchers with access to atomic force microscopes to perform scanning near field ultrasound holography to image nanoparticle interactions with cells. Over the course of this effort, grantees were able to develop a standalone module capable of performing scanning near field ultrasound holography on a variety of commercial systems. The add-on module was tested on systems from 3 of the major Atomic Force Microscope manufacturers (Veeco, JEOL and Park Systems). Furthermore, the investigators believe that the hardware developed will be capable of performing the technique on any AFM with standard signal access modules. The discrete components previously used to perform the SNFUH technique could cost upward of $\$$40,000. Beyond the cost issue, the components are bulky and cumbersome to work with because of the wiring needed between each of the components when implementing the technique. NanoSonix was able to dramatically reduce both the cost and complexity of implementing SNFUH, developing two functional versions of an add-on module (A&B) using National Instruments hardware as put forth in the proposal. More significantly, NanoSonix exceeded the aims of the proposal by developing completely custom electronics that should enable an almost order of magnitude cost reduction while improving the capability versus the discrete componentry. Because potential biological systems are so complex, biological applications of the technique are very diverse and require very distinct domain knowledge. This inhibits the utilization widespread adoption and penetration of new metrology techniques. Northwestern University and NanoSonix has made it both possible and practical for widespread adoption of the technique through the development of a standalone system with order of magnitude cost reduction and significant reduction in complexity. NanoSonix believes that it can produce the add-on modules and make them available to the academic research community at a price point below $\$$10,000, thereby enabling the biological research community to hardness the over 30,000 AFMs in existence around the world today, approximately 18,000 of which reside in the United States at a very modest incremental cost. Finally, we were able to implement a multi-frequency function generator and simultaneous multi-frequency lock-in amplifier with an internal reference frequency (which significantly simplifies the experimental setup) that it believes could lay the foundation for future non-destructive nanoscale 3-D tomography of buried and embedded sub-structures in cells and tissues. To date, we has been unable to find multiple frequency combinations which yield sufficiently strong mixing at the tip-sample interaction, but believes it is conceptually possible and doing so would enable the use of frequency dependent phase shift to better characterize the subsurface structures. Taken together, we believes that this work has materially advanced the prospects of enabling researchers to develop time-resolved cellular uptake of the nanoparticles used for drug delivery, cancer therapies, gene delivery and gene silencing and to assess cellular uptake of nanoparticles for environmental monitoring/remediation/impact.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

An Ontology Design Pattern for Surface Water Features

Surface water is a primary concept of human experience but concepts are captured in cultures and languages in many different ways. Still, many commonalities can be found due to the physical basis of many of the properties and categories. An abstract ontology of surface water features based only on those physical properties of landscape features has the best potential for serving as a foundational domain ontology. It can then be used to systematically incor-porate concepts that are specific to a culture, language, or scientific domain. The Surface Water ontology design pattern was developed both for domain knowledge distillation and to serve as a conceptual building-block for more complex surface water ontologies. A fundamental distinction is made in this on-tology between landscape features that act as containers (e.g., stream channels, basins) and the bodies of water (e.g., rivers, lakes) that occupy those containers. Concave (container) landforms semantics are specified in a Dry module and the semantics of contained bodies of water in a Wet module. The pattern is imple-mented in OWL, but Description Logic axioms and a detailed explanation is provided. The OWL ontology will be an important contribution to Semantic Web vocabulary for annotating surface water feature datasets. A discussion about why there is a need to complement the pattern with other ontologies, es-pecially the previously developed Surface Network pattern is also provided. Fi-nally, the practical value of the pattern in semantic querying of surface water datasets is illustrated through a few queries and annotated geospatial datasets.

GIS↗

3D face analysis for demographic biometrics

Despite being increasingly easy to acquire, 3D data is rarely used for face-based biometrics applications beyond identification. Recent work in image-based demographic biometrics has enjoyed much success, but these approaches suffer from the well-known limitations of 2D representations, particularly variations in illumination, texture, and pose, as well as a fundamental inability to describe 3D shape. This paper shows that simple 3D shape features in a face-based coordinate system are capable of representing many biometric attributes without problem-specific models or specialized domain knowledge. The same feature vector achieves impressive results for problems as diverse as age estimation, gender classification, and race classification.

Tokola, Ryan A↗

Research Opportunities in the Physical Design Optimization of Hybrid Power Plants: Preprint

With renewable energy growing to 10%-20% or more of overall electricity generation, design objectives for renewable power plants are shifting from producing energy at the lowest levelized cost to maximizing profitability. Several research and commercial efforts have investigated how to size respective generation and storage assets together for hybrid power plant systems - including wind, solar, battery storage and other technologies. However, the physical design of hybrid power plants has yet to be considered in detail. Even the design of individual solar and wind power plants is complex with large numbers of design variables, constraints and complex physical interactions across subsystems. Developing an integrated approach for the design of hybrid power plants including multiple generation technologies compounds this complexity and requires domain knowledge across the technologies as well as new approaches for addressing the statistical representation of resource availability and revenue opportunities. To inform the research needs on physical design of these hybrid power plants, a workshop was held in December 2018 at the National Renewable Energy Laboratory. This paper provides a summary of findings from the workshop on research needs for hybrid power plant design.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Graph Neural Networks for Particle Reconstruction in High Energy Physics detectors

Pattern recognition problems in high energy physics are notably different from traditional machine learning applications in computer vision. Reconstruction algorithms identify and measure the kinematic properties of particles produced in high energy collisions and recorded with complex detector systems. Two critical applications are the reconstruction of charged particle trajectories in tracking detectors and the reconstruction of particle showers in calorimeters. These two problems have unique challenges and characteristics, but both have high dimensionality, high degree of sparsity, and complex geometric layouts. Graph Neural Networks (GNNs) are a relatively new class of deep learning architectures which can deal with such data effectively, allowing scientists to incorporate domain knowledge in a graph structure and learn powerful representations leveraging that structure to identify patterns of interest. In this work we demonstrate the applicability of GNNs to these two diverse particle reconstruction problems.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

The CYBER security – Competency Health and Maturity Progression (CYBER-CHAMP) model: Extending the National Initiative for Cybersecurity Education (NICE) Framework Across Organizational Security

Problem Statement: There is a pervasive talent deficit in the cybersecurity industry that prevents employers from being able to fill their open positions efficiently. A holistic approach to security is required to ensure organizations have adequate prevention and response capabilities in case of a cyberattack. Specifically, industrial control systems (ICS’s) and their operational technology (OT) components have become a constant target for cyberattacks. Research Questions: It is proposed that the NICE Framework should be extended in the following areas: 1) Include guidance regarding the job roles and competencies for both IT and OT professionals. 2) Offer step-by-step solutions, based on the work role mappings from the NICE Framework, to increase cybersecurity through employee training and education. 3) Provide a streamlined, lifecycle approach to building a cybersecurity program. Contribution: The CYBER security – Competency Health and Maturity Progression (CYBER-CHAMP©) model provides a customized solution for businesses to understand their education gaps in organizational security and target areas for improvement. Rationale: The Framework for Improving Critical Infrastructure Cybersecurity v1.1 addresses ICS but does not offer a measurement of cybersecurity maturity or clear methods to ascertain an organization’s current risk profile. In Phases 1 and 5 of the model, measurements are provided to help an organization build their current and target risk profiles. The NICE framework provides a structure for planning an IT cybersecurity workforce, but the OT aspects of cybersecurity are only briefly discussed. The model uses Phases 2-3 to examine the competencies of an organization’s workforce, which includes both IT and OT roles. Current frameworks do not offer next steps to increase an organization’s cybersecurity. During Phase 4, employees’ roles are mapped to training, education, and/or certifications from common vendors. Investigative Approach: The model provides measurements and metrics for both an organization’s status and continual improvement. This improvement methodology includes guidance for creating an overall strategic plan for security improvement via products designed to increase an organization’s operational readiness through workforce competency health. Lessons Learned: Depending on who was participating, there were contradicting answers given in Phase 1 due to different security cultures in the organization. This revelation has influenced the steps listed in the User’s Guide, where Phase 1’s first recommended step is to assemble a team that champions the facilitation and implementation of the model in the organization. During Phase 2, the discovery was made that organizations may be missing roles that are necessary to perform critical cybersecurity functions. By understanding the functional roles and competencies needed, they can contract or hire cybersecurity help to fill these gaps. Implications: Using the model, organizations can discuss quantitative measures for improvement as a business case for advancing their security program. Future research can validate and extend the present theory and model to a variety of environments. It is of interest to investigate additional security roles and knowledge domains that are used to build standardized cybersecurity curriculum.

97 MATHEMATICS AND COMPUTING↗

Snowmass Computational Frontier: Topical Group Report on Quantum Computing

Quantum computing will play a pivotal role in the High Energy Physics (HEP) science program over the early parts of the 21$^{st}$ Century, both as a major expansion of our capabilities across the Computational Frontier, and in synthesis with quantum sensing and quantum networks. This report outlines how Quantum Information Science (QIS) and HEP are deeply intertwined endeavors that benefit enormously from a strong engagement together. Quantum computers do not represent a detour for HEP, rather they are set to become an integral part of our discovery toolkit. Problems ranging from simulating quantum field theories, to fully leveraging the most sensitive sensor suites for new particle searches, and even data analysis will run into limiting bottlenecks if constrained to our current computing paradigms. Easy access to quantum computers is needed to build a deeper understanding of these opportunities. In turn, HEP brings crucial expertise to the national quantum ecosystem in quantum domain knowledge, superconducting technology, cryogenic and fast microelectronics, and massive-scale project management. The role of quantum technologies across the entire economy is expected to grow rapidly over the next decade, so it is important to establish the role of HEP in the efforts surrounding QIS. Fully delivering on the promise of quantum technologies in the HEP science program requires robust support. It is important to both invest in the co-design opportunities afforded by the broader quantum computing ecosystem and leverage HEP strengths with the goal of designing quantum computers tailored to HEP science.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Real-Time Health Monitoring for Gas Turbine Components Using Online Learning and High-Dimensional Data (Final Report)

Capital-intensive turbomachinery, such as gas turbines and combined cycle plants, are constantly being monitored for performance anomalies, faults, and physical degradation. Although these power-generating assets are equipped with hundreds of sensors, existing monitoring tools can only handle moderate-sized data. As a result, only a handful of aggregate metrics are used to monitor machine health. At the same time, developing advanced tools suitable for large datasets have been restricted by the lack of appropriate data. The objective of this proposal was to demonstrate a Big Data analytics framework for fault detection and diagnosis in gas turbine applications. We develop a predictive analytics framework methodology guided by these experimental data, industrial data from our collaborators, and physics-based models with engineering domain knowledge. Our analytics framework consists of four key components: (1) a data curation process that addresses data storage, data quality assessments, and integrity checks, (2) a feature engineering component that utilizes statistical methods and transformation algorithms guided by physics-based models to extract high-fidelity fault features that can be leveraged for fault detection and classifying fault severities, (3) a Machine Learning-based fault detection and diagnostics algorithms for detecting operational and hardware faults in the combustion and the turbines section. We utilize two industry-class gas turbine component test rigs to generate first of its kind data for critical gas turbine faults with varying severity levels. Advanced gas turbine test facilities will be interrogated using state-of-the-art instrumentation techniques to build fault signatures and data trends for key combustor and turbine faults. Data generated from a combustor test rig (Georgia Tech) and a turbine test rig (Penn State) during both normal operation and with seeded faults serve as the basis for the Big Data sets. The test conditions in the two test facilities include common, critical events that occur in the operation. Utilizing the combustor test rig, we examine two common combustor faults: lean blowout and centerbody degradation. For the turbine section we develop analytic models for monitoring cooling faults in the gas turbine.

20 FOSSIL-FUELED POWER PLANTS↗

Lithium-Ion Battery Diagnostics Using Electrochemical Impedance via Machine-Learning

Diagnosing battery states such as health, state-of-charge, or temperature is crucial for ensuring the safety and reliability of electrochemical energy storage systems. While some states, such as temperature, may be measured using cheap sensors, accurate diagnosis of battery health metrics usually requires time-consuming performance measurements, making them infeasible for use in real-world operation. These health metrics can be measured during lab-testing and then estimated on-line using predictive life models or via state observer algorithms such as Kalman filters, but these predictive methods should be supplemented by actual measurement of battery health whenever possible to ensure reliability. Rapid measurement of battery health may be done by various types of fast diagnostic techniques such as electrochemical impedance spectroscopy (EIS), which can be performed in only a few minutes and require only a fraction of the energy and power needed for a full charge and discharge measurement. But there is a substantial challenge for estimating battery health using EIS data, as EIS is sensitive to cell temperature, state-of-charge, current, and resting time in addition to health. Thus, utilizing EIS data to predict battery capacity requires correcting for all these additional variables, a task that is extremely difficult to handle analytically. This talk utilizes machine-learning methods to estimate the effectiveness of battery capacity prediction from EIS data, leveraging a data set of hundreds of EIS measurements recorded at varying temperature and state-of-charge throughout a 500-day aging study of 32 commercial, large-format NMC-Graphite lithium-ion batteries. Using EIS as input to machine-learning models is complicated by the nonlinear response of impedance to battery health, temperature, and state-of-charge, as well as the collinearity between the impedance response at neighboring frequencies, which can easily lead to overfit models. To train robust models, features from EIS data need to be extracted from the data or some subset of critical frequencies selected. Many approaches for extracting and selecting features from EIS data from electrochemical analysis and machine-learning fields were identified for analysis: using the entire raw spectra; selection of one, two, or many frequencies from the entire spectra; selecting interesting points from the EIS measurement using domain knowledge; fitting EIS with an equivalent-circuit model; calculating statistics on the raw impedance values; and reducing the dimensionality of the data using unsupervised linear (principal component analysis) and non-linear (uniform manifold approximation and projection) methods. These approaches were rigorously compared using a machine-learning pipeline approach, training linear, Gaussian process, and random forest regression models and quantifying performance using cross-validation as well as a held-out test set. An artificial neural network model trained on the raw spectra was also tested. Promising pipelines were fine-tuned via Bayesian hyperparameter optimization using cross-validation loss and training with class-specific weights to counter data set imbalance. The most reliable method for utilizing impedance in this work was the selection of two optimal frequencies through an exhaustive search, resulting in about 2% mean absolute error on test data for both Gaussian process and random forest model architectures. Interrogation of a variety of models reveals critical frequencies of 100 Hz and 103 Hz for this data set, though the optimal set of frequencies is not necessarily intuitive, i.e., the best performing models are not simply those that use impedance at frequencies that have the highest correlation to the relative discharge capacity. The best performing model is an ensemble model, which is able to predict battery capacity with 1.9% mean absolute error for unseen cells using impedance recorded at a variety of temperatures and states-of-charge.

battery↗

Micro Baselines for Operational Technology Environments

Critical infrastructure stakeholders need to baseline their networks to understand expected communications. Top-down approaches to baselining rely on observables that are generally available but lack properties upon which traditional statistical tools depend. We propose to construct micro-baselines: signatures within operational networks based on observables associated with specific events. Such observables are informed by precursor analysis reports of historical cyber attacks on operational environments developed by Cybersecurity for Operational Technology Environments (CyOTE). Baseline measurements depend upon context beyond the cyber domain. An energy plant's baseline running in the summer may statistically differ from a similar facility in a colder region. Domain knowledge must be integrated to apply general micro-baselining algorithms to a facility-specific context. Therefore, we propose to explore the feasibility of transferring micro baselining algorithms across different facilities. Facilities that implement the same processes in different geographic locations will be compared relative to observable measurements used in micro-baselining for comparable events. One evaluation approach would condition or augment dynamic observables measured within a facility network testbed with additional observables derived from geographic context or infrastructure dependencies such as those provided by the All-Hazards Analysis tool.

97 MATHEMATICS AND COMPUTING↗

NuGraph2: A Graph Neural Network for Neutrino Event Reconstruction

Neutrino experiments are set to probe some of the most important open questions in physics, from CP violation and the nature of dark matter. The technology of choice for many of these experiments is the liquid argon time projection chamber (LArTPC). In current LArTPC experiments, reconstruction performance often represents a limiting factor for the sensitivity. New developments are therefore needed to unlock the full potential of LArTPC experiments. NuGraph2 is a state of the art Graph Neural Network for reconstruction of data in LArTPC experiments. NuGraph2 utilizes a heterogeneous graph structure, with separate subgraphs of 2D nodes (hits in each plane) connected across planes via 3D nodes (space points). The model provides a consistent description of the neutrino interaction across all planes. NuGraph2 is a multi-purpose network, with a common message-passing attention engine connected to multiple decoders with different classification or regression tasks. These include the classification of detector hits according to the particle type that produced them (semantic segmentation) and the separation of hits from the neutrino interaction from hits due to noise or cosmic-ray background. Additional decoders are being developed, performing tasks such as the regression of the neutrino interaction vertex position. Performance results will be presented based on publicly available samples from MicroBooNE. These include both physics performance metrics, achieving 95% accuracy for semantic segmentation and 98% classification of neutrino hits, as well as computational metrics for training and for inference on CPU or GPU. The status of the NuGraph integration in the LArSoft software framework will be presented, as well as initial studies about model interpretability and injection of domain knowledge.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

NuGraph2: A Graph Neural Network for Neutrino Event Reconstruction

Neutrino experiments are set to probe some of the most important open questions in physics, from CP violation and the nature of dark matter. The technology of choice for many of these experiments is the liquid argon time projection chamber (LArTPC). In current LArTPC experiments, reconstruction performance often represents a limiting factor for the sensitivity. New developments are therefore needed to unlock the full potential of LArTPC experiments. NuGraph2 is a state of the art Graph Neural Network for reconstruction of data in LArTPC experiments [https://arxiv.org/abs/2403.11872]. NuGraph2 utilizes a heterogeneous graph structure, with separate subgraphs of 2D nodes (hits in each plane) connected across planes via 3D nodes (space points). The model provides a consistent description of the neutrino interaction across all planes. NuGraph2 is a multi-purpose network, with a common message-passing attention engine connected to multiple decoders with different classification or regression tasks. These include the classification of detector hits according to the particle type that produced them (semantic segmentation) and the separation of hits from the neutrino interaction from hits due to noise or cosmic-ray background. Additional decoders are being developed, performing tasks such as the regression of the neutrino interaction vertex position. Performance results will be presented based on publicly available samples from MicroBooNE. These include both physics performance metrics, achieving 95% accuracy for semantic segmentation and 98% classification of neutrino hits, as well as computational metrics for training and for inference on CPU or GPU. The status of the NuGraph integration in the LArSoft software framework will be presented, as well as initial studies about model interpretability and injection of domain knowledge.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗