Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Deep Neural Networks”

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 1,423 records · Page 79

Domain Adaptation for Measurements of Strong Gravitational Lenses

Upcoming surveys are predicted to discover galaxy-scale strong lenses on the magnitude of 105, making deep learning methods necessary in lensing data analysis. Currently, there is insufficient real lensing data to train deep learning algorithms, but training only on simulated data results in poor performance on real data. Domain adaptation can bridge the gap between simulated and real datasets. We adopt domain adaptation on the estimation of Einstein radius in simulated galaxy-scale gravitational lensing images. We evaluate two domain adaptation techniques - domain adversarial neural networks (DANN) and maximum mean discrepancy (MMD). We train on a source domain of simulated lenses and apply it to a target domain with emulation of DES survey conditions. We show that both domain adaptation techniques can significantly improve the model performance on the more complex target domain datasets. Our results show the potential of using domain adaptation to perform analysis on future survey data with a deep neural network trained on simulated data.

79 ASTRONOMY AND ASTROPHYSICS↗

A Deep Learning Modeling Framework to Capture Mixing Patterns in Reactive-Transport Systems

Prediction and control of chemical mixing are vital for many scientific areas such as subsurface reactive transport, climate modeling, combustion, epidemiology, and pharmacology. Due to the complex nature of mixing in heterogeneous and anisotropic media, the mathematical models related to this phenomenon are not analytically tractable. Numerical simulations often provide a viable route to predict chemical mixing accurately. However, contemporary modeling approaches for mixing cannot utilize available spatial-temporal data to improve the accuracy of the future prediction and can be compute-intensive, especially when the spatial domain is large and for long-term temporal predictions. To address this knowledge gap, in this work we will present in this paper a deep learning (DL) modeling framework applied to predict the progress of chemical mixing under fast bimolecular reactions. This framework uses convolutional neural networks (CNN) for capturing spatial patterns and long short-term memory (LSTM) networks for forecasting temporal variations in mixing. By careful design of the framework—placement of non-negative constraint on the weights of the CNN and the selection of activation function, the framework ensures non-negativity of the chemical species at all spatial points and for all times. Our DL-based framework is fast, accurate, and requires minimal data for training. The time needed to obtain a forecast using the model is a fraction (≈ O(-6)) of the time needed to obtain the result using a high-fidelity simulation. To achieve an error of 10% (measured using the infinity norm) for capturing local-scale mixing features such as interfacial mixing, only 24% to 32% of the sequence data for model training is required. To achieve the same level of accuracy for capturing global-scale mixing features, the sequence data required for model training is 64% to 70% of the total spatial-temporal data. Hence, the proposed approach—a fast and accurate way to forecast long-time spatial-temporal mixing patterns in heterogeneous and anisotropic media—will be a valuable tool for modeling reactive-transport in a wide range of applications.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

PERSIANN-Unet: A Global Deep Learning Framework for Near-Real-Time Precipitation Estimation Using Infrared Data

Access to high-quality, high-resolution, near-real-time precipitation data is essential for hydrological and meteorological research and disaster mitigation. Traditional tools such as rain gauges and radar networks, though effective, have limitations, including sparse coverage in remote areas and high operational costs. Satellite data, with its global coverage and high spatial and temporal resolutions, mitigates limitations in coverage. Satellite precipitation products like Hydro Estimator (HE), Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (IMERG), and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) utilize both geosynchronous thermal infrared (IR) and passive microwave (PMW) data in their operation. PMW sensors offer detailed atmospheric profiles but suffer from higher latency, whereas IR sensors provide lower latency but only capture cloud-top information. Despite this constraint, IR data remains attractive for low-latency precipitation estimation. Recent advances in deep learning, particularly convolutional neural networks (CNNs), have further improved satellite precipitation retrievals. This study introduces PERSIANN-Unet (PUnet or PERSIANN V3), a quasi-global algorithm covering 60°N–60°S that combines IR data, monthly climatology, and the UNet architecture to produce half-hourly precipitation estimates at 0.04° resolution. The product is evaluated against HE, IMERG, and PDIR-Now for 2022–2023. Results show that PUnet closely matches its training target, IMERG V07 Final, at the global scale, and performance is further evaluated against Stage IV as a reference over CONUS. Training PUnet on IMERG (2016–2021) leverages a high-quality, integrated PMW IR-gauge precipitation product while developing an IR-based framework not reliant on PMW availability. By operating on a single global image, PUnet avoids tile partitioning and blending steps, reducing edge discontinuities, and produces more spatially consistent precipitation fields across hemispheres.

Phu Nguyen↗

Inferring safety critical events from vehicle kinematics in naturalistic driving environment: Application of deep learning Algorithms

Advances in sensing technology has enabled the collection of countless terabytes of second-by-second kinematics data. Such data provides opportunities for real-time monitoring of driving behavior and identification of safety critical events (SCEs) including crashes and near crashes. The concept of volatility is relevant in this context, which identifies instability and erratic variations in driving behavior prior to involvement in SCEs. This study utilized vehicle kinematics from a large-scale naturalistic driving data to develop a deep learning approach based on 1D convolutional neural networks (CNN) for inferring SCEs. The data are unique in the sense that such accurate pre-crash data at high fidelity are not available in traditional crash repositories. This study contributes to the literature by providing a first attempt at predicting responses to SCEs by developing deep learning-based CNN architectures using novel driving volatility based kinematic thresholds for a sample of 9553 events. The key contribution lies in developing a volatility-based CNN input layout that is acceptable to CNN schemes and represents the motion kinematics such as speed, acceleration and volatility measures. Several 1D-CNN architectures were developed using layers, numbers of convolutions, layer patterns, and kernels. Shallow and deep architectures were tested, revealing higher accuracy of shallow architectures in detecting SCEs. The optimal number of epochs were identified using an early stopping method while the CNN performance was improved by increasing the number of epochs. The ensemble CNN had the highest predictive accuracy of 95.6% for detection of crashes and near crashes, which was 2.5% higher than the optimal CNN using 20% hold out test data. The ensemble CNN also outperformed classical machine learning models and model performance reported in past studies on detection of SCEs. Finally, these results have implications for identification of safety hotspots and providing real-time alerts and warnings in connected and highly automated vehicle environment including society of automotive engineers levels 3–5.

42 ENGINEERING↗

Domain Adaptation for Measurements of Strong Gravitational Lenses

Upcoming surveys are predicted to discover galaxy-scale strong lenses on the order of 10\textsuperscript{5}, making deep learning methods necessary in lensing data analysis. Currently, there is insufficient real lensing data to train deep learning algorithms, but the alternative of training only on simulated data results in poor performance on real data. Domain Adaptation may be able to bridge the gap between simulated and real datasets. We utilize domain adaptation for the estimation of Einstein radius ($\Theta_E$) in simulated galaxy-scale gravitational lensing images with different levels of observational realism. We evaluate two domain adaptation techniques - Domain Adversarial Neural Networks (DANN) and Maximum Mean Discrepancy (MMD). We train on a source domain of simulated lenses and apply it to a target domain of lenses simulated to emulate noise conditions in the Dark Energy Survey (DES). We show that both domain adaptation techniques can significantly improve the model performance on the more complex target domain dataset. This work is the first application of domain adaptation for a regression task in strong lensing imaging analysis. Our results show the potential of using domain adaptation to perform analysis of future survey data with a deep neural network trained on simulated data.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Domain Adaptation for Measurements of Strong Gravitational Lenses

Upcoming surveys are predicted to discover galaxy-scale strong lenses on the order of $10^5$, making deep learning methods necessary in lensing data analysis. Currently, there is insufficient real lensing data to train deep learning algorithms, but the alternative of training only on simulated data results in poor performance on real data. Domain Adaptation may be able to bridge the gap between simulated and real datasets. We utilize domain adaptation for the estimation of Einstein radius ($\Theta_E$) in simulated galaxy-scale gravitational lensing images with different levels of observational realism. We evaluate two domain adaptation techniques - Domain Adversarial Neural Networks (DANN) and Maximum Mean Discrepancy (MMD). We train on a source domain of simulated lenses and apply it to a target domain of lenses simulated to emulate noise conditions in the Dark Energy Survey (DES). We show that both domain adaptation techniques can significantly improve the model performance on the more complex target domain dataset. This work is the first application of domain adaptation for a regression task in strong lensing imaging analysis. Our results show the potential of using domain adaptation to perform analysis of future survey data with a deep neural network trained on simulated data.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Generalization of Deep-Learning Models for Classification of Local Distance Earthquakes and Explosions across Various Geologic Settings

Although accurately classifying signals from earthquakes and explosions at local distance (<250 km) remains an important task for seismic network operations, the growing volume of available seismic data presents a challenge for analysts using traditional source discrimination techniques. In recent years, deep-learning models have proven effective at discriminating between low-magnitude earthquakes and explosions measured at local distances, but it is not clear how well these models are capable of generalizing across different geological settings. To address the issue of generalization between regions, we train deep-learning models (convolutional neural networks [CNNs]) on time–frequency representations (scalograms) of three-component earthquake and explosion signals from eight different regions in the continental United States. We explore scenarios where models are trained on data from all regions, individual regions, or all but one region. We find that although CNN models trained on individual regions do not necessarily generalize well across different settings, models trained on multiple regions that include diverse path coverage generalize to new regions, with station-level accuracy of up to 90% or more for data sets from unseen regions. In general, CNN-based discrimination models significantly outperform models based on uncorrected P/S ratio (measured in the 10–18 Hz frequency band), even when CNN models are tested on data from entirely unseen regions.

58 GEOSCIENCES↗

Even Higher-Level Synthesis: An Exploration of AI Hardware Accelerators using HLS4ML

With the rise of artificial intelligence, the popularization of deep learning, and a constantly evolving industry, the demand for flexible and efficient tools has never been greater. As algorithms grow more complex, their runtime and energy consumption increase exponentially. Customized hardware accelerators, long used for specific mathematical operations, remain essential for managing modern applications' computational and power demands. Hardware accelerators can speed up complex computations by orders of magnitude, but their manual design and verification processes are often challenging and time-consuming. High-Level Synthesis (HLS) provides a solution by transforming high-level algorithm descriptions, typically written in C++ or SystemC, into synthesizable RTL suitable for hardware implementation. This approach reduces development time for RTL engineers while offering flexibility beyond what traditional handwritten RTL can provide. We extended this capability to the machine-learning domain with the open-source framework hls4ml, which allows neural networks trained in Python frameworks like Tensorflow or PyTorch to be synthesized into efficient hardware representations for the traditional FPGA and ASIC flows. This breakthrough addresses the growing need for reduced design turnaround and easy verification of ML hardware accelerators with low latency and power efficiency constraints. During this tutorial, we will demonstrate how Python complements HLS by simplifying the ML design process, bridging the gap between software and hardware development. Attendees will explore how we translate neural networks modeled in Python into fixed-point C++ models suitable for HLS workflows. We will dive into strategies like Value-Range Analysis and Quantization-Aware Training, which optimize these designs for deployment and evaluate their accuracy, power consumption, and energy efficiency. To exemplify these concepts, experts from Fermilab will share their experiences applying this technology to high-energy physics experiments, where real-time, low-latency processing is critical. Over the years, Fermilab engineers have demonstrated how deep neural networks, optimized for hardware using hls4ml, can meet the stringent requirements of trigger systems at the CERN Large Hadron Collider. These systems rely on rapid decision-making to process immense data volumes while retaining only the most relevant events for further analysis. The application of hls4ml has also been extended to innovative technologies like smart pixel arrays. These smart pixels integrate ML inference capabilities directly into sensor devices, enabling localized data processing at the pixel level. This approach drastically reduces the need to transmit raw data to external processing units, significantly decreasing power consumption and latency. By embedding neural networks within the pixel architecture, the smart pixels can identify and prioritize relevant data in real time, providing a highly efficient solution for edge computing in scenarios such as particle detectors and imaging systems. Fermilab's work highlights the potential of hardware-accelerated ML in scenarios where both speed and power efficiency are mission-critical. Through this tutorial, attendees will gain valuable insights into the challenges and solutions of deploying ML in hardware. Understanding how HLS and hls4ml streamline the development of neural network-based hardware accelerators is fundamental for the industry's future. Participants will learn how these technologies are shaping the future of AI and scientific computing.

Di Guglielmo, Giuseppe [Fermilab]↗

Convolutional Neural Networks for the CHIPS Neutrino Detector R&D Project.

The CHerenkov detectors In mine PitS (Chips) neutrino detector R&D project aims to develop novel strategies and technologies for very large yet ‘cheap as chips’ water Cherenkov neutrino detectors. Via deployment in a body of water, use of commercially available components, and instrumentation coverage optimisation for the study of exclusively accelerator beam neutrinos, Chips will enable megaton scale detectors to become a reality at the cost of $200k-$300k per kt of sensitive mass. During the summer of 2019 a prototype Chips detector, Chips-5, was deployed into the Wentworth 2W disused mine pit in northern Minnesota, 7 mrad off the NuMI beam axis. A novel data acquisition system was introduced using cheap single-board computers and open-source software. This work presents a novel approach to water Cherenkov neutrino detector event reconstruction and classification. Three forms of a Convolutional Neural Network, a type of deep learning algorithm, have been trained to reject cosmic muon events, classify beam events, and estimate neutrino energies, all using only the raw detector event as input. When evaluated on the expected distribution of Chips-5 events, this new approach is shown to be robust and explainable as well as providing a significant performance increase over the standard likelihood-based reconstruction and simple neural network classification. Promisingly, the performance presented here is comparable to the more complex (and expensive) neutrino oscillation experiments within the field.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Photometric redshifts from SDSS images with an interpretable deep capsule network

ABSTRACT Studies of cosmology, galaxy evolution, and astronomical transients with current and next-generation wide-field imaging surveys like the Rubin Observatory Legacy Survey of Space and Time are all critically dependent on estimates of photometric redshifts. Capsule networks are a new type of neural network architecture that is better suited for identifying morphological features of the input images than traditional convolutional neural networks. We use a deep capsule network trained on ugriz images, spectroscopic redshifts, and Galaxy Zoo spiral/elliptical classifications of ∼400 000 Sloan Digital Sky Survey galaxies to do photometric redshift estimation. We achieve a photometric redshift prediction accuracy and a fraction of catastrophic outliers that are comparable to or better than current methods for SDSS main galaxy sample-like data sets (r ≤ 17.8 and zspec ≤ 0.4) while requiring less data and fewer trainable parameters. Furthermore, the decision-making of our capsule network is much more easily interpretable as capsules act as a low-dimensional encoding of the image. When the capsules are projected on a two-dimensional manifold, they form a single redshift sequence with the fraction of spirals in a region exhibiting a gradient roughly perpendicular to the redshift sequence. We perturb encodings of real galaxy images in this low-dimensional space to create synthetic galaxy images that demonstrate the image properties (e.g. size, orientation, and surface brightness) encoded by each dimension. We also measure correlations between galaxy properties (e.g. magnitudes, colours, and stellar mass) and each capsule dimension. We publicly release our code, estimated redshifts, and additional catalogues at https://biprateep.github.io/encapZulate-1.

79 ASTRONOMY AND ASTROPHYSICS↗

Automated correlative segmentation of large Transmission X-ray Microscopy (TXM) tomograms using deep learning

Highlights: • A unique automated segmentation approach for large 3D nanotomography datasets obtained by Transmission X-ray Microscopy (TXM) in an Al-Cu alloy. • Automated segmentation using a Convolutional Neural Network (CNN) architecture based on a deep learning approach was employed. • This extremely versatile technique is capable of emulating the manual segmentation process effectively. • Quantitative comparison between manual segmentation and CNN architecture established the accuracy of this automated technique. • Its ability to robustly process ultra-large volumes of data in relatively small time frames can exponentially accelerate tomographic data analysis. - Abstract: A unique correlative approach for automated segmentation of large 3D nanotomography datasets obtained using Transmission X-ray Microscopy (TXM) in an Al-Cu alloy has been introduced. Automated segmentation using a Convolutional Neural Network (CNN) architecture based on a deep learning approach was employed. This extremely versatile technique is capable of emulating the manual segmentation process effectively. Coupling this technique with post-scanning SEM imaging ensured precise estimation of 3D morphological parameters from nanotomography. The segmentation process as well as subsequent analysis was expedited by several orders of magnitude. Quantitative comparison between segmentation performed manually and using the CNN architecture established the accuracy of this automated technique. Its ability to robustly process ultra-large volumes of data in relatively small time frames can exponentially accelerate tomographic data analysis, possibly opening up novel avenues for performing 4D characterization experiments with finer time steps.

36 MATERIALS SCIENCE↗

Ramifications of Evolving Misbehaving Convolutional Neural Network Kernel and Batch Sizes

Deep-learners have many hyper-parameters including learning rate, batch size, kernel size - all playing a significant role toward estimating high quality models. Discovering useful hyper-parameter guidelines is an active area of research, though the state of the art generally uses a brute force, uniform grid approach or random search for finding ideal settings. We share the preliminary results of using an alternative approach to deep learner hyper-parameter tuning that uses an evolutionary algorithm to improve the accuracy of a deep-learner models used in satellite imagery building footprint detection. We found that the kernel and batch size hyper-parameters surprisingly differed from sizes arrived at via a brute force uniform grid approach. These differences suggest a novel role for evolutionary algorithms in determining the number of convolution layers, as well as smaller batch sizes in improving deep-learner models.

Coletti, Mark↗

A Two-Tier Convolutional Neural Network for Combined Detection and Segmentation in Biological Imagery

Deep learning techniques have been useful for modern microscopy imaging techniques to further study and analyze biological structures and organs. Convolutional neural networks (CNN) have improved 2D object detection, localization, and segmentation. For imagery containing biological structures with depth, it is especially desirable to perform these tasks in 3D. Traditionally, performing these tasks simultaneously in 3D has proven to be computationally expensive. Currently available methodologies thus largely work to segment 3D objects from 2D images (without context from captured 3D volumes). In this work, we present a novel approach to perform fast and accurate localization, detection, and segmentation of volumes containing cells. Specifically, in our method, we modify and tune two state-of-the-art CNNs, namely 2D YOLOv2 and 3D U-Net, and combine them with a new fusion and image processing algorithms. Annotated volumes in this space are limited, and we have created synthetic data that mimics actual structures for training and testing our proposed approach. Promising results on this test data demonstrate the value of the technique and offers a methodology for 3D cell analysis in real microscopy imagery.

Ziabari, Amir↗

Deeply Uncertain: Comparing Methods of Uncertainty Quantification in Deep Learning Algorithms [Slides]

Deep learning is used in many applications in the physical sciences. In those sciences we are used to having uncertainty on every measurement or prediction. In the last years, many methods have been put forth for uncertainty quantification: Bayesian Neural Networks, Concrete Dropout, and Deep Ensembles. Uncertainty in deep learning is often divided into aleatoric or irreducible and epistemic or reducible. The problem remains: Which uncertainty quantification method should be chosen? How are those results interpreted?

97 MATHEMATICS AND COMPUTING↗

Machine Learning Predictions of Transition Probabilities in Atomic Spectra

Forward modeling of optical spectra with absolute radiometric intensities requires knowledge of the individual transition probabilities for every transition in the spectrum. In many cases, these transition probabilities, or Einstein A-coefficients, quickly become practically impossible to obtain through either theoretical or experimental methods. Complicated electronic orbitals with higher order effects will reduce the accuracy of theoretical models. Experimental measurements can be prohibitively expensive and are rarely comprehensive due to physical constraints and sheer volume of required measurements. Due to these limitations, spectral predictions for many element transitions are not attainable. In this work, we investigate the efficacy of using machine learning models, specifically fully connected neural networks (FCNN), to predict Einstein A-coefficients using data from the NIST Atomic Spectra Database. For simple elements where closed form quantum calculations are possible, the data-driven modeling workflow performs well but can still have lower precision than theoretical calculations. For more complicated nuclei, deep learning emerged more comparable to theoretical predictions, such as Hartree–Fock. Unlike experiment or theory, the deep learning approach scales favorably with the number of transitions in a spectrum, especially if the transition probabilities are distributed across a wide range of values. It is also capable of being trained on both theoretical and experimental values simultaneously. In addition, the model performance improves when training on multiple elements prior to testing. The scalability of the machine learning approach makes it a potentially promising technique for estimating transition probabilities in previously inaccessible regions of the spectral and thermal domains on a significantly reduced timeline.

74 ATOMIC AND MOLECULAR PHYSICS↗

Deep Learning for Fish Identification from Sonar Data: CRADA 481 [Abstract only]

To help solve the challenges of hydropower energy production related to the potential for eel injury and mortality from passage through hydropower turbines, we will develop a deep learning method for identifying migrating eels from imaging sonar. This project continues with a prior project conducted by the Pacific Northwest National Laboratory (PNNL) and the Electric Power Research Institute (EPRI) in FY2018-2019. The proposed method employs Convolution Neural Network (CNN), a powerful deep learning method for image classification, to distinguish between images of eels and non-eel moving objects. We propose to collect more laboratory data and add more existing field data to train a powerful deep learning model. In addition to eels and sticks as classified in previous studies, we will add images containing several non-eel fish species and macrophyte mats to the training data. A multi-class classification model will be developed to distinguish these objects. Object detection algorithm will be explored and developed to locate and identify multiple objects in each sonar frame. Motion analysis will be performed to track the movement of objects in sonar video clips. We will also improve the data conversion algorithm so that it can read in both DIDSON and ARIS (both are imaging sonars developed by Sound Metrics Corp) data files and convert them to images with comparably high resolution, regardless of the varying detection ranges in different environments. The developed algorithms will be packaged as a software with a graphic user interface. The software will be evaluated by external collaborators in the field. The developed framework can be generalized for automatic monitoring of fish passage and migration using other imaging sonars like ARIS and will benefit the design and operation of ecologically friendly hydroelectric projects. The developed wavelet and CNN model configuration parameters can potentially be transferred to lamprey detection in similar riverine environments.

13 HYDRO ENERGY↗

Progress Toward a Deep-Learning-Based Search for Low-Energy Electron Neutrinos in MicroBooNE

MicroBooNE is a neutrino experiment based at Fermilab which consists of a liquid argon time-projection chamber in the Booster Neutrino Beam (BNB). The experiment aims to investigate the excess of electron neutrino-like events seen by the MiniBooNE experiment, also located in the BNB. I will describe the status of our deep-learning-based search for low-energy electron neutrino interactions within the MicroBooNE detector and the muon neutrino interactions by which we will constrain them. This analysis features a novel hybrid approach of traditional reconstruction methods combined with the use of convolutional neural networks, a type of deep learning algorithm highly adept at pattern recognition. I will present work on ongoing studies that will characterize and quantify the systematic uncertainties associated with this analysis.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗