Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “matching pursuit”

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 19 records
All sourcesNASA NTRSDOE OSTI
NASA NTRS2009

Highly Scalable Matching Pursuit Signal Decomposition Algorithm

Matching Pursuit Decomposition (MPD) is a powerful iterative algorithm for signal decomposition and feature extraction. MPD decomposes any signal into linear combinations of its dictionary elements or atoms . A best fit atom from an arbitrarily defined dictionary is determined through cross-correlation. The selected atom is subtracted from the signal and this procedure is repeated on the residual in the subsequent iterations until a stopping criterion is met. The reconstructed signal reveals the waveform structure of the original signal. However, a sufficiently large dictionary is required for an accurate reconstruction; this in return increases the computational burden of the algorithm, thus limiting its applicability and level of adoption. The purpose of this research is to improve the scalability and performance of the classical MPD algorithm. Correlation thresholds were defined to prune insignificant atoms from the dictionary. The Coarse-Fine Grids and Multiple Atom Extraction techniques were proposed to decrease the computational burden of the algorithm. The Coarse-Fine Grids method enabled the approximation and refinement of the parameters for the best fit atom. The ability to extract multiple atoms within a single iteration enhanced the effectiveness and efficiency of each iteration. These improvements were implemented to produce an improved Matching Pursuit Decomposition algorithm entitled MPD++. Disparate signal decomposition applications may require a particular emphasis of accuracy or computational efficiency. The prominence of the key signal features required for the proper signal classification dictates the level of accuracy necessary in the decomposition. The MPD++ algorithm may be easily adapted to accommodate the imposed requirements. Certain feature extraction applications may require rapid signal decomposition. The full potential of MPD++ may be utilized to produce incredible performance gains while extracting only slightly less energy than the standard algorithm. When the utmost accuracy must be achieved, the modified algorithm extracts atoms more conservatively but still exhibits computational gains over classical MPD. The MPD++ algorithm was demonstrated using an over-complete dictionary on real life data. Computational times were reduced by factors of 1.9 and 44 for the emphases of accuracy and performance, respectively. The modified algorithm extracted similar amounts of energy compared to classical MPD. The degree of the improvement in computational time depends on the complexity of the data, the initialization parameters, and the breadth of the dictionary. The results of the research confirm that the three modifications successfully improved the scalability and computational efficiency of the MPD algorithm. Correlation Thresholding decreased the time complexity by reducing the dictionary size. Multiple Atom Extraction also reduced the time complexity by decreasing the number of iterations required for a stopping criterion to be reached. The Course-Fine Grids technique enabled complicated atoms with numerous variable parameters to be effectively represented in the dictionary. Due to the nature of the three proposed modifications, they are capable of being stacked and have cumulative effects on the reduction of the time complexity.

Christensen, Daniel↗
NASA NTRS1994

Optical Implementation of Matching Pursuit for Image Representation

We have developed a technique for image analysis, representation, and decomposition. This technique was motivated by Stephane Mallat's matching-pursuit algorithm. We've altered and simplified the mechanics of his algorithm to enable an extremely fast implementation via optical processing. Initial computer simulations show that our algorithm is capable of decomposing and representing a 2-D image as a linear combination of basis images with both high speed and high fidelity.

matching pursuit adaptive wavelet transform image ↗
NASA NTRS1996

Use of the Matching Pursuit Algorithm for Flight Flutter Test Data Analysis

The goal of flight flutter testing is to detect possibly destructive modes of aircraft vibration which may arise during flight from interaction of aerodynamic forces with structural dynamic properties of the airframe. This is typically accomplished by exciting the airframe with a time varying force and monitoring the response of the aircraft throughout its flight envelope. The data generated must be analyzed and presented so that the frequency and time of occurrence of excited modes are clearly and unambiguously displayed. Processing and display in near real time is also desirable. Display of data in the time-frequency plane is a natural choice because it is a familiar and intuitive framework. The Matching Pursuit algorithm provides a time-frequency analysis with good adaptability to signal structure and good signal representation in the time-frequency plane. Improvements in efficiency are needed before the algorithm can be used in real time, however.

Nelson, Ronald E.↗
NASA NTRS1998

Dictionary Approaches to Image Compression and Reconstruction

This paper proposes using a collection of parameterized waveforms, known as a dictionary, for the purpose of medical image compression. These waveforms, denoted as lambda, are discrete time signals, where y represents the dictionary index. A dictionary with a collection of these waveforms Is typically complete or over complete. Given such a dictionary, the goal is to obtain a representation Image based on the dictionary. We examine the effectiveness of applying Basis Pursuit (BP), Best Orthogonal Basis (BOB), Matching Pursuits (MP), and the Method of Frames (MOF) methods for the compression of digitized radiological images with a wavelet-packet dictionary. The performance of these algorithms is studied for medical images with and without additive noise.

Ziyad, Nigel A.↗
NASA NTRS1998

Dictionary Approaches to Image Compression and Reconstruction

This paper proposes using a collection of parameterized waveforms, known as a dictionary, for the purpose of medical image compression. These waveforms, denoted as phi(sub gamma), are discrete time signals, where gamma represents the dictionary index. A dictionary with a collection of these waveforms is typically complete or overcomplete. Given such a dictionary, the goal is to obtain a representation image based on the dictionary. We examine the effectiveness of applying Basis Pursuit (BP), Best Orthogonal Basis (BOB), Matching Pursuits (MP), and the Method of Frames (MOF) methods for the compression of digitized radiological images with a wavelet-packet dictionary. The performance of these algorithms is studied for medical images with and without additive noise.

Ziyad, Nigel A.↗
DOE OSTI2020

Path-Based Dictionary Augmentation: A Framework for Improving $k$ -Sparse Image Processing

In this study, we have previously shown that augmenting orthogonal matching pursuit (OMP) with an additional step in the identification stage of each pursuit iteration yields improved $k$ -sparse reconstruction and denoising performance relative to baseline OMP. At each iteration a “path” or geodesic, is generated between the two dictionary atoms that are most correlated with the residual and from this path a new atom that has a greater correlation to the residual than either of the two bracketing atoms is selected. Here, we provide new computational results illustrating improvements in sparse coding and denoising on canonical datasets using both learned and structured dictionaries. The two methods of constructing a path are investigated for each dictionary type: the Euclidean geodesic formed by a linear combination of the two atoms and the 2-Wasserstein geodesic corresponding to the optimal transport map between the atoms. We prove here the existence of a higher-correlation atom in the Euclidean case under assumptions on the two bracketing atoms and introduce algorithmic modifications to improve the likelihood that the bracketing atoms meet those conditions. Although, we demonstrate our augmentation on OMP alone, in general it may be applied to any reconstruction algorithm that relies on the selection and sorting of high-similarity atoms during an analysis or identification phase.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Investigating explainable transfer learning for battery lifetime prediction under state transitions

Battery lifetime prediction at early cycles is crucial for researchers and manufacturers to examine product quality and promote technology development. Machine learning has been widely utilized to construct data-driven solutions for high-accuracy predictions. However, the internal mechanisms of batteries are sensitive to many factors, such as charging/discharging protocols, manufacturing/storage conditions, and usage patterns. These factors will induce state transitions, thereby decreasing the prediction accuracy of data-driven approaches. Transfer learning is a promising technique that overcomes this difficulty and achieves accurate predictions by jointly utilizing information from various sources. Hence, we develop two transfer learning methods, Bayesian Model Fusion and Weighted Orthogonal Matching Pursuit, to strategically combine prior knowledge with limited information from the target dataset to achieve superior prediction performance. From our results, our transfer learning methods reduce root-mean-squared error by 41% through adapting to the target domain. Furthermore, the transfer learning strategies identify the variations of impactful features across different sets of batteries and therefore disentangle the battery degradation mechanisms and the root cause of state transitions from the perspective of data mining. These findings suggest that the transfer learning strategies proposed in our work are capable of acquiring knowledge across multiple data sources for solving specialized issues.

25 ENERGY STORAGE↗
DOE OSTI2022

GenMod: A generative modeling approach for spectral representation of PDEs with random inputs

Here, we propose a method for quantifying uncertainty in high-dimensional PDE systems with random parameters, where the number of solution evaluations is small. Parametric PDE solutions are often approximated using a spectral decomposition based on polynomial chaos expansions. For the class of systems we consider (i.e., high dimensional with limited solution evaluations) the coefficients are given by an underdetermined linear system in a regression formulation. This implies additional assumptions, such as sparsity of the coefficient vector, are needed to approximate the solution. Here, we present an approach where we assume the coefficients are close to the range of a generative model that maps from a low to a high dimensional space of coefficients. Our approach is inspired be recent work examining how generative models can be used for compressed sensing in systems with random Gaussian measurement matrices. Using results from PDE theory on coefficient decay rates, we construct an explicit generative model that predicts the polynomial chaos coefficient magnitudes. The algorithm we developed to find the coefficients, which we call GenMod, is composed of two main steps. First, we predict the coefficient signs using Orthogonal Matching Pursuit. Then, we assume the coefficients are within a sparse deviation from the range of a sign-adjusted generative model. This allows us to find the coefficients by solving a nonconvex optimization problem, over the input space of the generative model and the space of sparse vectors. We obtain theoretical recovery results for a Lipschitz continuous generative model and for a more specific generative model, based on coefficient decay rate bounds. We examine three high-dimensional problems and show that, for all three examples, the generative model approach outperforms sparsity promoting methods at small sample sizes.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1997

Estimation of Modal Parameters Using a Wavelet-Based Approach

Modal stability parameters are extracted directly from aeroservoelastic flight test data by decomposition of accelerometer response signals into time-frequency atoms. Logarithmic sweeps and sinusoidal pulses are used to generate DAST closed loop excitation data. Novel wavelets constructed to extract modal damping and frequency explicitly from the data are introduced. The so-called Haley and Laplace wavelets are used to track time-varying modal damping and frequency in a matching pursuit algorithm. Estimation of the trend to aeroservoelastic instability is demonstrated successfully from analysis of the DAST data.

Lind, Rick↗
NASA NTRS2008

Sparse Regression as a Sparse Eigenvalue Problem

We extend the l0-norm "subspectral" algorithms for sparse-LDA [5] and sparse-PCA [6] to general quadratic costs such as MSE in linear (kernel) regression. The resulting "Sparse Least Squares" (SLS) problem is also NP-hard, by way of its equivalence to a rank-1 sparse eigenvalue problem (e.g., binary sparse-LDA [7]). Specifically, for a general quadratic cost we use a highly-efficient technique for direct eigenvalue computation using partitioned matrix inverses which leads to dramatic x103 speed-ups over standard eigenvalue decomposition. This increased efficiency mitigates the O(n4) scaling behaviour that up to now has limited the previous algorithms' utility for high-dimensional learning problems. Moreover, the new computation prioritizes the role of the less-myopic backward elimination stage which becomes more efficient than forward selection. Similarly, branch-and-bound search for Exact Sparse Least Squares (ESLS) also benefits from partitioned matrix inverse techniques. Our Greedy Sparse Least Squares (GSLS) generalizes Natarajan's algorithm [9] also known as Order-Recursive Matching Pursuit (ORMP). Specifically, the forward half of GSLS is exactly equivalent to ORMP but more efficient. By including the backward pass, which only doubles the computation, we can achieve lower MSE than ORMP. Experimental comparisons to the state-of-the-art LARS algorithm [3] show forward-GSLS is faster, more accurate and more flexible in terms of choice of regularization

Exact Sparse Least Squares (ESLS)↗
NASA NTRS2023

Differential Saccade-Pursuit Coordination Under Sleep Loss and Low-Dose Alcohol

Introduction: Ocular tracking of a moving object requires tight coordination between smooth pursuit and saccadic eye movements. Normally, pursuit drives gaze velocity to closely match target velocity, with residual position offsets corrected by catch-up saccades. However, how/if common stressors affect this coordination is largely unknown. This study seeks to elucidate the effects of acute and chronic sleep loss, and low-dose alcohol, on saccade-pursuit coordination, as well as that of caffeine. Methods: We used an ocular tracking paradigm to assess three metrics of tracking (pursuit gain, saccade rate, saccade amplitude) and to compute “ground lost” (from reductions in steady-state pursuit gain) and “ground recouped” (from increases in steady-state saccade rate and/or amplitude). We emphasize that these are measures of relative changes in positional offsets, and not absolute offset from the fovea. Results: Under low-dose alcohol and acute sleep loss, ground lost was similarly large. However, under the former, it was nearly completely recouped by saccades, whereas under the latter, compensation was at best partial. Under chronic sleep restriction and acute sleep loss with a caffeine countermeasure, the pursuit deficit was dramatically smaller, yet saccadic behavior remained altered from baseline. In particular, saccadic rate remained significantly elevated, despite the fact that ground lost was minimal. Discussion: This constellation of findings demonstrates differential impacts on saccade-pursuit coordination with low-dose alcohol impacting only pursuit, likely through extrastriate cortical pathways, while acute sleep loss not only disrupts pursuit but also undermines saccadic compensation, likely through midbrain/brainstem pathways. Furthermore, while chronic sleep loss and caffeine-mitigated acute sleep loss show little residual pursuit deficit, consistent with uncompromised cortical visual processing, they nonetheless show an elevated saccade rate, suggesting residual midbrain and/or brainstem impacts.

smooth pursuit↗
NASA NTRS2024

Oculometric Biomarkers of Visuomotor Deficits in Clinically Asymptomatic Patients With Systemic Lupus Erythematosus Undergoing Long-Term Hydroxychloroquine Treatment

Introduction: This study examines a set of oculomotor measurements, or “oculometric” biomarkers, as potential early indicators of visual and visuomotor deficits due to retinal toxicity in asymptomatic Systemic Lupus Erythematosus (SLE) patients on long-term hydroxychloroquine (HCQ) treatment. The aim is to identify subclinical functional impairments that are otherwise undetectable by standard clinical tests and to link them to structural retinal changes. Methods: We measured oculomotor responses in a cohort of SLE patients on chronic HCQ therapy using a previously established behavioral task and analysis technique. We also examined the relationship between oculometrics, OCT measures of retinal thickness, and standard clinical perimetry measures of visual function in our patient group using Bivariate Pearson Correlation and a Linear Mixed-Effects Model (LMM). Results: Significant visual and visuomotor deficits were found in 12 asymptomatic SLE patients on long-term HCQ therapy compared to a cohort of 17 age-matched healthy controls. Notably, six oculometrics were significantly different. The median initial pursuit acceleration was 22%, steady-state pursuit gain 16%, proportion smooth 7%, and target speed responsiveness 31% lower, while catch-up saccade amplitude was 46% and fixation error 46% larger. Excluding the two patients with diagnosed mild toxicity, four oculometrics, all but fixation error and proportion smooth, remained significantly impaired compared to controls. Across our population of 12 patients (24 retinae), we found that pursuit latency, initial acceleration, steady-state gain, and fixation error were linearly related to retinal thickness even when age was accounted for, while standard measures of clinical function (Mean Deviation and Pattern Standard Deviation) were not. Discussion: Our data show that specific oculometrics are sensitive early biomarkers of functional deficits in SLE patients on HCQ that could be harnessed to assist in the early detection of HCQ-induced retinal toxicity and other visual pathologies, potentially providing early diagnostic value beyond standard visual field and OCT evaluations.

eye movements↗
DOE OSTI2020

AWCC Simulations

Neutron well counters provide a means to measure fissile mass by detecting coincident neutrons, which is a unique signature to fission. For coincidence counting, mass is determined using a calibration curve established with items that have the same material characteristics. Multiplicity analysis solves the point model equations for mass using known detector parameters. In both cases, mass measurements may exhibit a bias due to item characteristics and should be corrected for. Experiments can be done to estimate bias and material effects, but they require having both the well counter and the material well defined. In pursuit of estimating measurement bias and uncertainty in a High Efficiency Neutron Counter (HENC), an MCNP simulation was matched to measurements performed with an Active Well Coincidence Counter (AWCC) to de ne a neutron source term. The neutron source term will be used to predict measurement performance of similar material types in a HENC.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
NASA NTRS1998

Minimization of Retinal Slip Cannot Explain Human Smooth-Pursuit Eye Movements

Existing models assume that pursuit attempts a direct minimization of retinal image motion or "slip" (e.g. Robinson et al., 1986; Krauzlis & Weisberger, 1989). Using occluded line-figure stimuli, we have previously shown that humans can accurately pursue stimuli for which perfect tracking does not zero retinal slip (Neurologic ARCO). These findings are inconsistent with the standard control strategy of matching eye motion to a target-motion signal reconstructed by adding retinal slip and eye motion, but consistent with a visual front-end which estimates target motion via a global spatio-temporal integration for pursuit and perception. Another possible explanation is that pursuit simply attempts to minimize slip perpendicular to the segments (and neglects parallel "sliding" motion). To resolve this, 4 observers (3 naive) were asked to pursue the center of 2 types of stimuli with identical velocity-space descriptions and matched motion energy. The line-figure "diamond" stimulus was viewed through 2 invisible 3 deg-wide vertical apertures (38 cd/m2 equal to background) such that only the sinusoidal motion of 4 oblique line segments (44 cd/m2 was visible. The "cross" was identical except that the segments exchanged positions. Two trajectories (8's and infinity's) with 4 possible initial directions were randomly interleaved (1.25 cycles, 2.5s period, Ax = Ay = 1.4 deg). In 91% of trials, the diamond appeared rigid. Correspondingly, pursuit was vigorous (mean Again: 0.74) with a V/H aspect ratio approx. 1 (mean: 0.9). Despite a valid rigid solution, the cross however appeared rigid in 8% of trials. Correspondingly, pursuit was weaker (mean Hgain: 0.38) with an incorrect aspect ratio (mean: 1.5). If pursuit were just minimizing perpendicular slip, performance would be the same in both conditions.

Stone, Leland S.↗
DOE OSTI2021

Minimum Resolution Requirements for Gamma Identification Algorithms

Each year there are millions of dollars spent on the research and production of high-resolution detectors. This research indicates that the pursuit of higher resolution detectors is not always necessary. The terminal resolution of a NaI detector, or highest detector resolution, at which identification algorithms fail to identify highly enriched uranium (HEU) was evaluated using GADRAS, Genie, and GammaVision. GADRAS employs a template matching algorithm, while Genie and GammaVision utilize a mathematical approach for peak search and identification. The NaI spectra utilized for evaluation were generated using the GADRAS Inject tab and source modeling functions. Each spectrum included terrestrial and cosmic background from Dallas, TX. The resolutions for each spectrum were increased from a default 8.92% to a point where each algorithm would fail to identify 235 U from a HEU source. Six different source configurations were used in this research: bare HEU, 50% shielded HEU, 90% shielded HEU, bare HEU with an interference source of 99 mTc, bare HEU with 99 mTc both shielded 50%, and bare HEU with 99 mTc both shielded 90%. T

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2024

Validation of Local Structural Loads Computed by OpenFAST Against Measurements From the FOCAL Experimental Campaign

This work presents the validation of the local structural load modeling capability in OpenFAST for floating substructures based on data from the FOCAL experimental campaign. Previously, OpenFAST could only represent the floating substructure as a rigid body, and though this approach can model the global response of the floater in most cases, it is not able to capture the structural loads within the floater's individual members. Consideration of local substructure loads is important for some floating designs, because the pursuit of cost reduction often results in lighter and more flexible structures. To address this limitation, the HydroDyn (hydrodynamics) and SubDyn (substructure dynamics) modules of OpenFAST have been recently extended to account for the flexibility of floating substructures. To validate this new capability, we compare the results obtained by OpenFAST with data measured during the FOCAL experimental campaign, which analyzed a 1:70 scale performance-matched model of the IEA 15-MW reference turbine atop a modified University of Maine VolturnUS-S semisubmersible in a wave basin under the action of both wind and waves. For the purposes of the present work, the most important feature of the experiment is the presence of load cells at the root of each pontoon, and our objective is to assess how well those loads are reproduced by OpenFAST. To model the distributed hydrodynamic and hydrostatic loads along the floating substructure, we adopt a strip-theory approach based on the Morison equation, and we discuss the impact of different hydrodynamic modeling options (wave stretching, MacCamy-Fuchs correction, and second-order wave kinematics) on both motions and loads. For simplicity, we focus on wave-only conditions, both regular and irregular. The results demonstrate good overall agreement for the loads at the root of the pontoons for the waves analyzed in this work, especially given the assumptions and simplifications inherent to a simple strip-theory model.

floating offshore wind turbine↗
DOE OSTI2024

Validation of Local Structural Loads Computed by OpenFAST Against Measurements From the Focal Experimental Campaign: Preprint

This work presents the validation of the local structural load modeling capability in OpenFAST for floating substructures based on data from the FOCAL experimental campaign. Previously, OpenFAST could only represent the floating substructure as a rigid body, and though this approach can model the global response of the floater in most cases, it is not able to capture the structural loads within its individual members. Consideration of local substructure loads is important for some floating designs, as the pursuit of cost reduction often results in lighter and more flexible structures. To address this limitation, the HydroDyn (hydrodynamics) and SubDyn (substructure dynamics) modules of OpenFAST have been recently extended to account for the flexibility of floating substructures. To validate this new capability, we compare the results obtained by OpenFAST with data measured during the FOCAL experimental campaign, which analyzed a 1:70 scale performance-matched model of the IEA 15-MW reference turbine atop a modified University of Maine VolturnUS-S semisubmersible in a wave basin under the action of both wind and waves. For the purposes of the present work, the most important feature of the experiment is the presence of load cells at the root of each pontoon, and our objective is to assess how well those loads are reproduced by OpenFAST. To model the distributed hydrodynamic and hydrostatic loads along the floating substructure, we adopt a strip-theory approach based on the Morison equation, and we discuss the impact of different hydrodynamic modeling options (wave stretching, MacCamy-Fuchs correction, and second-order wave kinematics) on both motions and loads. For simplicity, we focus on wave-only conditions, both regular and irregular. The results demonstrate good overall agreement for the loads at the root of the pontoons for the waves analyzed in this work, especially given the assumptions and simplifications inherent to a simple strip-theory model.

floating offshore wind turbine↗