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 1,135 records · Page 63

A General Framework to Learn Tertiary Structure for Protein Sequence Characterization

During the past five years, deep-learning algorithms have enabled ground-breaking progress towards the prediction of tertiary structure from a protein sequence. Very recently, we developed SAdLSA, a new computational algorithm for protein sequence comparison via deep-learning of protein structural alignments. SAdLSA shows significant improvement over established sequence alignment methods. In this contribution, we show that SAdLSA provides a general machine-learning framework for structurally characterizing protein sequences. By aligning a protein sequence against itself, SAdLSA generates a fold distogram for the input sequence, including challenging cases whose structural folds were not present in the training set. About 70% of the predicted distograms are statistically significant. Although at present the accuracy of the intra-sequence distogram predicted by SAdLSA self-alignment is not as good as deep-learning algorithms specifically trained for distogram prediction, it is remarkable that the prediction of single protein structures is encoded by an algorithm that learns ensembles of pairwise structural comparisons, without being explicitly trained to recognize individual structural folds. As such, SAdLSA can not only predict protein folds for individual sequences, but also detects subtle, yet significant, structural relationships between multiple protein sequences using the same deep-learning neural network. The former reduces to a special case in this general framework for protein sequence annotation.

59 BASIC BIOLOGICAL SCIENCES↗

Microparticle cloud imaging and tracking for data-driven plasma science

Oceans of image and particle track data encountered in plasma interactions with microparticle clouds motivate development and applications of machine-learning (ML) algorithms. A local-constant-velocity tracker, a Kohonen neural network or self-organizing map, the feature tracking kit, and U-Net are described and compared with each other for microparticle cloud datasets generated from exploding wires, dusty plasmas, and atmospheric plasmas. Particle density and the signal-to-noise ratio have been identified as two important factors that affect the tracking accuracy. Fast Fourier transform is used to reveal how U-Net, a deep convolutional neural network developed for non-plasma applications, achieves the improvements for noisy scenes. Viscous effects are revealed in the ballistic motions of the particles from the exploding wires and atmospheric plasmas. Subdiffusion of microparticles satisfying Δr 2 ∝t k (k=0.84±0.02) is obtained from the dusty plasma datasets. Microparticle cloud imaging and tracking, when enhanced with data and ML models, present new possibilities for plasma physics.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Synthesis of CdZnTeSe single crystals for room temperature radiation detector fabrication: mitigation of hole trapping effects using a convolutional neural network

In this article, we report the growth of Cd 0.9 Zn 0.1 Te 0.97 Se 0.03 (CZTS) wide bandgap semiconductor single crystals for room temperature gamma-ray detection using a modified vertical Bridgman method. Charge transport properties measured in the radiation detectors, fabricated from the grown CZTS crystals, indicated signs of hole trapping. Hole traps inhibit high-resolution radiation detection especially for energetic gamma rays. Machine learning (ML) applications are gaining tremendous mpetus in improving device and sensor performance by compensating for limi tations arising from such intrinsic material properties. In this article, we describe a deep convolutional neural network (CNN) that has demonstrated remarkable efficiency in identifying the energy of a gamma photon detected by a CZTS detector. The CNN has been trained using simulated data that resemble output pulses from actual CZTS detectors when exposed to 662-keV gamma photons. The device properties required for the simulation have been derived from radiation detection measurements on a real Cd 0.9 Zn 0.1 Te 0.97 Se 0.03 detector fabricated in our laboratory. The CNN has been trained with detector pulses arising through photoelectric (PE) and Compton scattering (CS) separately. The percentage error in predicting the detected energies, within an extremely small duration of 0.28 ms, was found to be lower than 0.1% for gamma energies above 50 keV and for training datasets con taining PE and CS events separately. The CNN was also validated for a mixed PE and CS dataset to obtain a prediction error of 1%. Additionally, the effect of detector resolution on the efficiency of the CNN was also explored.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Exact enforcement of temporal continuity in sequential physics-informed neural networks

The use of deep learning methods in scientific computing represents a potential paradigm shift in engineering problem solving. One of the most prominent developments is Physics-Informed Neural Networks (PINNs), in which neural networks are trained to satisfy partial differential equations (PDEs). While this method shows promise, the standard version has been shown to struggle in accurately predicting the dynamic behavior of time-dependent problems. To address this challenge, methods have been proposed that decompose the time domain into multiple segments, employing a distinct neural network in each segment and directly incorporating continuity between them in the loss function of the minimization problem. In this work we introduce a method to exactly enforce continuity between successive time segments via a solution ansatz. This hard constrained sequential PINN (HCS-PINN) method is simple to implement and eliminates the need for any loss terms associated with temporal continuity. The method is tested for a number of benchmark problems involving both linear and non-linear PDEs. Examples include various first order time dependent problems in which traditional PINNs struggle, namely advection, Allen–Cahn, and Korteweg–de Vries equations. Furthermore, second and third order time-dependent problems are demonstrated via wave and Jerky dynamics examples, respectively. Notably, the Jerky dynamics problem is chaotic, making the problem especially sensitive to temporal accuracy. Finally, the numerical experiments conducted with the proposed method demonstrated superior convergence and accuracy over both traditional PINNs and the soft-constrained counterparts.

42 ENGINEERING↗

Deep Learning Analysis of Polaritonic Wave Images

Deep learning (DL) is an emerging analysis tool across the sciences and engineering. Encouraged by the successes of DL in revealing quantitative trends in massive imaging data, we applied this approach to nanoscale deeply subdiffractional images of propagating polaritonic waves in complex materials. Utilizing the convolutional neural network (CNN), we developed a practical protocol for the rapid regression of images that quantifies the wavelength and the quality factor of polaritonic waves. Using simulated near-field images as training data, the CNN can be made to simultaneously extract polaritonic characteristics and material parameters in a time scale that is at least 3 orders of magnitude faster than common fitting/processing procedures. The CNN-based analysis was validated by examining the experimental near-field images of charge-transfer plasmon polaritons at graphene/α-RuCl3 interfaces. Our work provides a general framework for extracting quantitative information from images generated with a variety of scanning probe methods.

97 MATHEMATICS AND COMPUTING↗

Measurement of Long Baseline Neutrino Oscillations and Improvements from Deep Learning

NOvA is a long-baseline neutrino oscillation experiment which measures the oscillation of muon neutrinos from the NuMI beam at Fermilab after they travel through the Earth for 810 km. In this dissertation I describe the operations and monitoring of the detectors which make it possible to record over 98% of the delivered neutrino beam. I also present reconstruction and identification techniques using deep convolutional neural networks (CNNs), which are applicable to multiple analyses. Lastly, I detail the oscillation analyses in the $\nu_{\mu} \to \nu_{\mu}$ and $\nu_{\mu} \to\nu_e$ channels on a dataset of 8:85 x $10^{20}$ protons on target, with emphasis on the application of deep learning, energy reconstruction, and treatment of systematic uncertainties....

Psihas, Fernanda↗

Fast automated analysis of strong gravitational lenses with convolutional neural networks

Quantifying image distortions caused by strong gravitational lensing—the formation of multiple images of distant sources due to the deflection of their light by the gravity of intervening structures—and estimating the corresponding matter distribution of these structures (the ‘gravitational lens’) has primarily been performed using maximum likelihood modelling of observations. Our procedure is typically time- and resource-consuming, requiring sophisticated lensing codes, several data preparation steps, and finding the maximum likelihood model parameters in a computationally expensive process with downhill optimizers. Accurate analysis of a single gravitational lens can take up to a few weeks and requires expert knowledge of the physical processes and methods involved. Tens of thousands of new lenses are expected to be discovered with the upcoming generation of ground and space surveys. We report the use of deep convolutional neural networks to estimate lensing parameters in an extremely fast and automated way, circumventing the difficulties that are faced by maximum likelihood methods. We also show that the removal of lens light can be made fast and automated using independent component analysis of multi-filter imaging data. Our networks can recover the parameters of the ‘singular isothermal ellipsoid’ density profile, which is commonly used to model strong lensing systems, with an accuracy comparable to the uncertainties of sophisticated models but about ten million times faster: 100 systems in approximately one second on a single graphics processing unit. These networks can provide a way for non-experts to obtain estimates of lensing parameters for large samples of data.

79 ASTRONOMY AND ASTROPHYSICS↗

Validation of Image-Based Neural Network Controllersthrough Adaptive Stress Testing

Neural networks have become state-of-the-art for computer vision problems because of their ability to efficiently model complex functions from large amounts of data. While neural networks can be shown to perform well empirically fora variety of tasks, their performance is difficult to guarantee.Neural network verification tools have been developed that can certify robustness with respect to a given input image; however,for neural network systems used in closed-loop controllers,robustness with respect to individual images does not address multi-step properties of the neural network controller and itsenvironment. Furthermore, neural network systems interacting in the physical world and using natural images are operating in a black-box environment, making formal verification in-tractable. This work combines the adaptive stress testing (AST)framework with neural network verification tools to search for the most likely sequence of image disturbances that cause the neural network controlled system to reach a failure. Anautonomous aircraft taxi application is presented, and results show that the AST method finds failures with more likely image disturbances than baseline methods. Further analysis of AST results revealed an explainable cause of the failure, giving insight into the problematic scenarios that should be addressed.

Adaptive Stress Testing, Marabou, Deep Neural Netw↗

Cy-Phy ADS: Cyber Physical Anomaly Detection Framework for EV Charging Systems

Today’s large-scale Electric Vehicle (EV) infrastructures are heavily dependent on information communication technologies to maintain their operation and to support communication within sub-system components as well as the outside world. These technologies are vulnerable to various cyber and physical threats. Timely identification and mitigation of these threats are critical for improving human safety, avoiding economic losses, and preventing catastrophic system failures. By addressing this, our work presents a ResNet Autoencoder (AE) based Cyber-Physical Anomaly Detection System (Cy-Phy ADS) for detecting anomalies in EV Controller Area Network (CAN) protocol communication. It consists of four main components: Cyber-Physical Feature Extractor, ResNet AE-based Anomaly Detection Framework, Cyber-Physical Health Metric (CPHM), and Visualization Dashboard. The presented framework was trained and tested using CAN data collected from the EV charging system testbed at the Idaho National Laboratory. The presented Cy-Phy ADS compared against six widely used unsupervised anomaly detection algorithms: One Class Support Vector Machine (OCSVM), Variational Autoencoder (VAE), LSTM Autoencoder (LSTM AE), Isolation Forest (IForest), Principle Component Analysis (PCA) and Local Outlier Factor (LOF). Here the presented approach showed the highest accuracy among the compared methods. Further, the proposed approach showed comparable performance in terms of precision, F1, and False positive rate. It also showed the lowest training and inference time compared to the neural network-based baseline algorithms compared against with. Additionally, the Cy-Phy ADS has advantages such as unsupervised training, the ability to provide a holistic metric for system health characterization, and non-linear feature extraction.

99 GENERAL AND MISCELLANEOUS↗

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↗

Image-based novel fault detection with deep learning classifiers using hierarchical labels

One important characteristic of modern fault classification systems is the ability to flag the system when faced with previously unseen fault types. This work considers the unknown fault detection capabilities of deep neural network-based fault classifiers. Specifically, we propose a methodology on how, when available, labels regarding the fault taxonomy can be used to increase unknown fault detection performance without sacrificing model performance. To achieve this, we propose to utilize soft label techniques to improve the state-of-the-art deep novel fault detection techniques during the training process and novel hierarchically consistent detection statistics for online novel fault detection. Lastly, we demonstrated increased detection performance on novel fault detection in inspection images from the hot steel rolling process, with results well replicated across multiple scenarios and baseline detection methods.

42 ENGINEERING↗

Convolutional neural network based non-iterative reconstruction for accelerating neutron tomography *

Abstract Neutron computed tomography (NCT), a 3D non-destructive characterization technique, is carried out at nuclear reactor or spallation neutron source-based user facilities. Because neutrons are not severely attenuated by heavy elements and are sensitive to light elements like hydrogen, neutron radiography and computed tomography offer a complementary contrast to x-ray CT conducted at a synchrotron user facility. However, compared to synchrotron x-ray CT, the acquisition time for an NCT scan can be orders of magnitude higher due to lower source flux, low detector efficiency and the need to collect a large number of projection images for a high-quality reconstruction when using conventional algorithms. As a result of the long scan times for NCT, the number and type of experiments that can be conducted at a user facility is severely restricted. Recently, several deep convolutional neural network (DCNN) based algorithms have been introduced in the context of accelerating CT scans that can enable high quality reconstructions from sparse-view data. In this paper, we introduce DCNN algorithms to obtain high-quality reconstructions from sparse-view and low signal-to-noise ratio NCT data-sets thereby enabling accelerated scans. Our method is based on the supervised learning strategy of training a DCNN to map a low-quality reconstruction from sparse-view data to a higher quality reconstruction. Specifically, we evaluate the performance of two popular DCNN architectures—one based on using patches for training and the other on using the full images for training. We observe that both the DCNN architectures offer improvements in performance over classical multi-layer perceptron as well as conventional CT reconstruction algorithms. Our results illustrate that the DCNN can be a powerful tool to obtain high-quality NCT reconstructions from sparse-view data thereby enabling accelerated NCT scans for increasing user-facility throughput or enabling high-resolution time-resolved NCT scans.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Deep active learning for classifying cancer pathology reports

Abstract Background Automated text classification has many important applications in the clinical setting; however, obtaining labelled data for training machine learning and deep learning models is often difficult and expensive. Active learning techniques may mitigate this challenge by reducing the amount of labelled data required to effectively train a model. In this study, we analyze the effectiveness of 11 active learning algorithms on classifying subsite and histology from cancer pathology reports using a Convolutional Neural Network as the text classification model. Results We compare the performance of each active learning strategy using two differently sized datasets and two different classification tasks. Our results show that on all tasks and dataset sizes, all active learning strategies except diversity-sampling strategies outperformed random sampling, i.e., no active learning. On our large dataset (15K initial labelled samples, adding 15K additional labelled samples each iteration of active learning), there was no clear winner between the different active learning strategies. On our small dataset (1K initial labelled samples, adding 1K additional labelled samples each iteration of active learning), marginal and ratio uncertainty sampling performed better than all other active learning techniques. We found that compared to random sampling, active learning strongly helps performance on rare classes by focusing on underrepresented classes. Conclusions Active learning can save annotation cost by helping human annotators efficiently and intelligently select which samples to label. Our results show that a dataset constructed using effective active learning techniques requires less than half the amount of labelled data to achieve the same performance as a dataset constructed using random sampling.

59 BASIC BIOLOGICAL SCIENCES↗

Completion design improvement using a deep convolutional network

Maximizing stimulated natural and hydraulic fracture network is one of the primary hydraulic fracturing concerns for economic production from a horizontal shale gas well. Geomechanical facies and preexisting fractures in each stage are identified based on similarities in formation characteristics to optimize the locations of perforation clusters. This often requires analyzing large volumes of drilling, Logging While Drilling (LWD) and Measurement While Drilling (MWD) data. In this paper, we develop a methodology that calculates the mechanical specific energy (MSE) using real-time drill string acceleration signals directly from its definition. High resolution vibration signals have been collected using a tri-axial accerlometer, which was an auxiliary tool included in acoustic borehole imager. This technique provides a cost-efficient solution for engineered completion design. Furthermore, we adopt deep Convolutional Neural Network (CNN) with signal processing to build a data pipeline that effectively extracts patterns from dynamic acceleration signals for rock lateral MSE classification. First, we apply discrete wavelet transform and Short-Time Fourier Transform (STFT) for signal denoising and pattern recognition. Then we construct an image dataset using multi-scale image fusion at pixel level from 3 sensor channels, including axial, lateral acceleration spectrograms and zero-padded revolutions per minute (RPM). The resulted RGB image dataset includes 4,000 images of 5 MSE ranges with various rock strength conditions. Our results demonstrate that the proposed deep learning model can achieve more than 90% classification accuracy. The deep learning results, as a reference source, were applied in selected Marcellus Shale Energy and Environmental Lab (MSEEL) wells engineered completion located in the Marcellus shale gas site.

03 NATURAL GAS↗

Neural network denoising of x-ray images from high-energy-density experiments

Noise is a consistent problem for x-ray transmission images of High-Energy-Density (HED) experiments because it can significantly affect the accuracy of inferring quantitative physical properties from these images. We consider experiments that use x-ray area backlighting to image a thin layer of opaque material within a physics package to observe its hydrodynamic evolution. The spatial variance of the x-ray transmission across the system due to changing opacity serves as an analog for measuring density in this evolving layer. The noise in these images adds nonphysical variations in measured intensity, which can significantly reduce the accuracy of our inferred densities, particularly at small spatial scales. Denoising these images is thus necessary to improve our quantitative analysis, but any denoising method also affects the underlying information in the image. In this paper, we present a method for denoising HED x-ray images via a deep convolutional neural network model with a modified DenseNet architecture. In our denoising framework, we estimate the noise present in the real (data) images of interest and apply the inferred noise distribution to a set of natural images. These synthetic noisy images are then used to train a neural network model to recognize and remove noise of that character. We show that our trained denoiser network significantly reduces the noise in our experimental images while retaining important physical features.

47 OTHER INSTRUMENTATION↗

Efficient proactive vehicle relocation for on-demand mobility service with recurrent neural networks

One major challenge for on-demand mobility service (OMS) providers is to seamlessly match empty vehicles with trip requests so that the total vacant mileage is minimized. In this work, we develop an innovative data-driven approach for devising efficient vehicle relocation policy for OMS that (1) proactively relocates vehicles before the demand is observed and (2) reduces the inequality among drivers' income so that the proactive relocation policy is fair and is likely to be followed by drivers. Our approach represents the fusion of optimization and machine learning methods, which comprises three steps: First, we formulate the optimal proactive relocation as an optimal/stable matching problems and solve for global optimal solutions based on historical data. Second, the optimal solutions are then grouped and fed to train the deep learning models which consist of fully connected layers and long short-term memory networks. Low rank approximation is introduced to reduce the model complexity and improve the training performances. Finally, we use the trained model to predict the relocation policy which can be implemented in real time. We conduct comprehensive numerical experiments and sensitivity analyses to demonstrate the performances of the proposed method using New York City taxi data. Here, the results suggest that our method will reduce empty mileage per trip by 54-70% under optimal matching strategy, and a 25-32% reduction can also be achieved by following stable matching strategy. We also validate that the predicted relocation policies are robust in the presence of uncertain passenger demand level and passenger trip-requesting behavior.

33 ADVANCED PROPULSION SYSTEMS↗

Rethinking the ill-posedness of the spectral function reconstruction — Why is it fundamentally hard and how Artificial Neural Networks can help

Reconstructing hadron spectral functions through Euclidean correlation functions are of the important missions in lattice QCD calculations. However, in a Källen–Lehmann (KL) spectral representation, the reconstruction is observed to be ill-posed in practice. It is usually ascribed to the fewer observation points compared to the number of points in the spectral function. In this paper, by solving the eigenvalue problem of continuous KL convolution, we show analytically that the ill-posedness of the inversion is fundamental and it exists even for continuous correlation functions. We discussed how to introduce regulators to alleviate the predicament, in which include the Artificial Neural Networks (ANNs) representations recently proposed by the Authors in another study. The uniqueness of solutions using ANNs representations is manifested analytically and validated numerically. Reconstructed spectral functions using different regularization schemes are also demonstrated, together with their eigen-mode decomposition. We observe that components with large eigenvalues can be reliably reconstructed by all methods, whereas those with low eigenvalues need to be constrained by regulators.

97 MATHEMATICS AND COMPUTING↗

LEMD

Deep neural network-based force field enables researchers to study materials systems of large size and/or time scale efficiently with comparable accuracy to DFT-based methods. However, such force fields, if improperly trained or mistakenly selected, could result in ill-conditioned results. This SBIR Phase I project has explored the possibility to characterize and index force fields in Deep potential theme by quantizing local atomic environments of atoms in a simulation system. This project has also demonstrated the feasibility of building a public accessible web infrastructure providing high quality machine learning algorithm-based force field for next generation materials modeling and simulation to accelerate the study and discovery of new energy-related materials.

Su, Ninghai↗