Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “image classification”

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 145 records · Page 8

Transfer Learning approach to parking lot classification in aerial imagery

The importance of satellite imagery analysis has increased dramatically over the last several years, keeping pace with the rapid improvements seen in both remote sensing platforms and sensors. As this field expands, so too does the interest in using machine learning methods to automate parts of the imagery analyst’s workflow. In this paper we address one aspect of this challenge: the development of a method for the automatic extraction of parking lots from aerial imagery. To the best of our knowledge, there has been no prior work conducted on the development of an end-to-end pipeline for this particular task. Due to the limited size of our dataset and to accommodate the potentially limited size of future datasets, we propose a deep learning approach using transfer learning. This process hinges upon the use of state of the art Convolutional Neural Networks (CNNs), trained on general image classification datasets. These networks were then fine-tuned on our custom dataset, to establish a comprehensive benchmark for this task. Our method exhibits promising results for automatic parking lot extraction, and is generalizable enough to work with different input types, including high resolution aerial orthoimagery, satellite imagery, full motion video (FMV), and UAV imagery.

97 MATHEMATICS AND COMPUTING↗

VoroClust: Scalable Clustering for Remote Sensing

Although supervised machine learning provides a powerful framework for image classification and segmentation, it requires comprehensive consistent datasets, which are not available for many remote-sensing applications. Remote-sensing datasets are expensive to collect, and each is acquired under different environmental conditions or with significant variations in system operating parameters. Unsupervised clustering algorithms analyze the structure of each dataset independently, rather than drawing on similarities with existing “training” examples, and are thus well suited for practical remote-sensing applications. We introduce VoroClust, a fast density-based unsupervised clustering algorithm applicable to high-resolution and high-dimensional data. VoroClust runs as fast as distance-based clustering methods, while capturing complex regional geometries at least as well as current-density-based methods. It uses a data-centered sphere cover to reduce computational demands, while still capturing data topology. It then propagates clusters outward from local peaks in density. We show that VoroClust provides fast state-of-the-art clustering for both high-resolution polarimetric synthetic aperture radar and high-dimensional hyperspectral imaging datasets.

42 ENGINEERING↗

Development of a Methodology to Assess the Impact of Non-Conventional Waste Streams Following a Radiological Incident - 18271

The Environmental Protection Agency's (EPA's) Waste Estimation Support Tool (WEST) is a Geographic Information System (GIS)-based decision support tool for estimating waste generated from remediation and cleanup activities following a radiological incident. WEST uses the Federal Emergency Management Agency's (FEMA's) Hazus-MH software's infrastructure information combined with image classification capabilities to estimate the quantity of waste and debris that may result from a radiological incident. WEST can be used to create customized decontamination strategies that allow end-users to evaluate different means of decontamination and how those strategies may impact the resulting waste stream in each geographical locality. This capability can potentially reduce the response/recovery time frame and cost. One of the key observations from past applications of WEST is that planning scenarios are site-specific, especially when dealing with chemical, biological, radiological, or nuclear (CBRN) waste management issues. Furthermore, site-specific issues are not limited to resource limitations or response bottlenecks but should also account for differences in an area's geography and urban footprint, a phenomenon observed during the Fukushima response. For instance, a recent WEST case study found that the amount of waste generated per unit of affected area, for a given decontamination approach, is largely a function of the building surface area in the affected area (relative to the ground surfaces). However, differences in the geometry and density of contaminated objects in the landscape may impact the amount and characteristics of the waste stream. Although buildings typically constitute a large portion of the surface area in urban environments, other objects such as vehicles and trees can be just as prominent. For example, seasonal changes in the amount of foliage present can alter the non-structural surface area calculated for an area like New York City by 20-30%. Very little is known about the impact that these non-conventional waste streams have on overall remediation strategies, or to what extent differences in characteristics of urban versus rural environments, traffic activities, or seasonal variations may have on these factors. This paper will present the proposed methodology to incorporate into WEST the impacts that niche, nonconventional waste streams might have from rural and urban environments following a nuclear power plant incident or detonation of a radiological dispersal device or improvised nuclear device. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Acute Superior Mesenteric Venous Thrombosis: Transcatheter Thrombolysis and Aspiration Thrombectomy Therapy by Combined Route of Superior Mesenteric Vein and Artery in Eight Patients

PurposeTo assess the feasibility, effectiveness, and safety of catheter-directed thrombolysis and aspiration thrombectomy therapy by combined route of superior mesenteric vein and artery (SMV+SMA) for acute superior mesenteric venous thrombosis (ASMVT).MethodsThis retrospective study reviewed eight ASMVT patients with transcatheter direct thrombolysis and aspiration thrombectomy therapy via SMV and indirect thrombolysis via SMA during a period of 14 months. The demographics, etiology, risk factors, therapeutic effect, complications, mortality, and follow-up of the study population were assessed. Anatomic and imaging classification of location and extent of thrombus at diagnosis and degree of thrombus lysis were described.ResultsTechnical success was achieved with substantial improvement in symptoms and thrombus resolution after thrombolytic therapy in all patients. The local urokinase infusion by SMA and SMV was performed for 5–7 (6.13 ± 0.83) and 7–15 (12 ± 2.51) days. Anticoagulation was performed catheter-directed and then orally throughout hospitalization and after discharge. Four patients required delayed localized bowel resection after thrombolytic therapy with no death. Thrombolytic therapy was not interrupted despite minor bleeding at the puncture site in two patients and sepsis in another two postoperatively. Nearly complete removal of thrombus was demonstrated by contrast-enhanced CT scan and portography before discharge. Patients were discharged in 10–27 (19.25 ± 4.89) days after admission. No recurrence developed during the follow-up of 10–13 (12.13 ± 0.99) months.ConclusionsCatheter-directed thrombolytic and aspiration therapy via SMV+SMA is beneficial for ASMVT in avoiding patient death, efficient resolving thrombus, rapid improving symptoms, reversing extensive intestinal ischemia, averting bowel resection, or localizing infarcted bowel segment and preventing short bowel syndrome.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

PlasmoData.jl — A Julia framework for modeling and analyzing complex data as graphs

Datasets encountered in scientific and engineering applications appear in complex formats (e.g., images, multivariate time series, molecules, video, text strings, networks). Graph theory provides a unifying framework to model such datasets and enables the use of powerful tools that can help analyze, visualize, and extract value from data. In this work, we present PlasmoData.jl, an open-source, Julia framework that uses concepts of graph theory to facilitate the modeling and analysis of complex datasets. The core of our framework is a general data modeling abstraction, which we call a DataGraph. We show how the abstraction and software implementation can be used to represent diverse data objects as graphs and to enable the use of tools from topology, graph theory, and machine learning (e.g., graph neural networks) to conduct a variety of tasks. We illustrate the versatility of the framework by using real datasets: (i) an image classification problem using topological data analysis to extract features from the graph model to train machine learning models; (ii) a disease outbreak problem where we model multivariate time series as graphs to detect abnormal events; and (iii) a technology pathway analysis problem where we highlight how we can use graphs to navigate connectivity. Further, our discussion also highlights how PlasmoData.jl leverages native Julia capabilities to enable compact syntax, scalable computations, and interfaces with diverse packages. Overall, we show that the DataGraph abstraction and PlasmoData.jl Julia package are able to model data within graphs and enable useful analysis.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

QoS-aware edge AI placement and scheduling with multiple implementations in FaaS-based edge computing

Resource constraints on the computing continuum require that we make smart decisions for serving AI-based services at the network edge. AI-based services typically have multiple implementations (e.g., image classification implementations include SqueezeNet, DenseNet, and others) with varying trade-offs (e.g., latency and accuracy). The question then is how should AI-based services be placed across Function-as-a-Service (FaaS) based edge computing systems in order to maximize total Quality-of-Service (QoS). To address this question, we propose a problem that jointly aims to solve (i) edge AI service placement and (ii) request scheduling. These are done across two time-scales (one for placement and one for scheduling). Here we first cast the problem as an integer linear program. We then decompose the problem into separate placement and scheduling subproblems and prove that both are NP-hard. We then propose a novel placement algorithm that places services while considering device-to-device communication across edge clouds to offload requests to one another. Our results show that the proposed placement algorithm is able to outperform a state-of-the-art placement algorithm for AI-based services, and other baseline heuristics, with regard to maximizing total QoS. Additionally, we present a federated learning-based framework, FLIES, to predict the future incoming service requests and their QoS requirements. Our results also show that our FLIES algorithm is able to outperform a standard decentralized learning baseline for predicting incoming requests and show comparable predictive performance when compared to centralized training.

97 MATHEMATICS AND COMPUTING↗

Confidence-weighted integration of human and machine judgments for superior decision-making

Large language models (LLMs) can surpass humans in certain forecasting tasks. What role does this leave for humans in the overall decision process? One possibility is that humans, despite performing worse than LLMs, can still add value when teamed with them. A human and machine team can surpass each individual teammate when team members’ confidence is well calibrated and team members diverge in which tasks they find difficult (i.e., calibration and diversity are needed). We simplified and extended a Bayesian approach to combining judgments using a logistic regression framework that integrates confidence-weighted judgments for any number of team members. Using this straightforward method, we demonstrated its effectiveness in both image classification and neuroscience forecasting tasks. Combining human judgments with one or more machines consistently improved overall team performance. Our hope is that this simple and effective strategy for integrating the judgments of humans and machines will lead to productive collaborations.

97 MATHEMATICS AND COMPUTING↗

$\mathrm{SIM}$-$\mathrm{STEM}$ $\mathrm{Lab}$: Incorporating Compressed Sensing Theory for Fast STEM Simulation

We report recently it has been shown that precise dose control and an increase in the overall acquisition speed of atomic resolution scanning transmission electron microscope (STEM) images can be achieved by acquiring only a small fraction of the pixels in the image experimentally and then reconstructing the full image using an inpainting algorithm. In this paper, we apply the same inpainting approach (a form of compressed sensing) to simulated, sub-sampled atomic resolution STEM images. We find that it is possible to significantly sub-sample the area that is simulated, the number of g-vectors contributing the image, and the number of frozen phonon configurations contributing to the final image while still producing an acceptable fit to a fully sampled simulation. Here we discuss the parameters that we use and how the resulting simulations can be quantifiably compared to the full simulations. As with any Compressed Sensing methodology, care must be taken to ensure that isolated events are not excluded from the process, but the observed increase in simulation speed provides significant opportunities for real time simulations, image classification and analytics to be performed as a supplement to experiments on a microscope to be developed in the future

47 OTHER INSTRUMENTATION↗

Electron Tomography and Machine Learning for Understanding the Highly Ordered Structure of Leafhopper Brochosomes

Insects known as leafhoppers (Hemiptera: Cicadellidae) produce hierarchically structured nanoparticles known as brochosomes that are exuded and applied to the insect cuticle, thereby providing camouflage and anti-wetting properties to aid insect survival. Although the physical properties of brochosomes are thought to depend on the leafhopper species, the structure–function relationships governing brochosome behavior are not fully understood. Brochosomes have complex hierarchical structures and morphological heterogeneity across species, due to which a multimodal characterization approach is required to effectively elucidate their nanoscale structure and properties. In this work, we study the structural and mechanical properties of brochosomes using a combination of atomic force microscopy (AFM), electron microscopy (EM), electron tomography, and machine learning (ML)-based quantification of large and complex scanning electron microscopy (SEM) image data sets. This suite of techniques allows for the characterization of internal and external brochosome structures, and ML-based image analysis methods of large data sets reveal correlations in the structure across several leafhopper species. Our results show that brochosomes are relatively rigid hollow spheres with characteristic dimensions and morphologies that depend on leafhopper species. Nanomechanical mapping AFM is used to determine a characteristic compression modulus for brochosomes on the order of 1–3 GPa, which is consistent with crystalline proteins. Altogether, this work provides an improved understanding of the structural and mechanical properties of leafhopper brochosomes using a new set of ML-based image classification tools that can be broadly applied to nanostructured biological materials.

Chemical structure↗

CryoEM reveals the stochastic nature of individual ATP binding events in a group II chaperonin

Chaperonins are homo- or hetero-oligomeric complexes that use ATP binding and hydrolysis to facilitate protein folding. ATP hydrolysis exhibits both positive and negative cooperativity. The mechanism by which chaperonins coordinate ATP utilization in their multiple subunits remains unclear. Here we use cryoEM to study ATP binding in the homo-oligomeric archaeal chaperonin from Methanococcus maripaludis (MmCpn), consisting of two stacked rings composed of eight identical subunits each. Using a series of image classification steps, we obtained different structural snapshots of individual chaperonins undergoing the nucleotide binding process. We identified nucleotide-bound and free states of individual subunits in each chaperonin, allowing us to determine the ATP occupancy state of each MmCpn particle. We observe distinctive tertiary and quaternary structures reflecting variations in nucleotide occupancy and subunit conformations in each chaperonin complex. Detailed analysis of the nucleotide distribution in each MmCpn complex indicates that individual ATP binding events occur in a statistically random manner for MmCpn, both within and across the rings. Our findings illustrate the power of cryoEM to characterize a biochemical property of multi-subunit ligand binding cooperativity at the individual particle level.

59 BASIC BIOLOGICAL SCIENCES↗

Metaplastic and energy-efficient biocompatible graphene artificial synaptic transistors for enhanced accuracy neuromorphic computing

CMOS-based computing systems that employ the von Neumann architecture are relatively limited when it comes to parallel data storage and processing. In contrast, the human brain is a living computational signal processing unit that operates with extreme parallelism and energy efficiency. Although numerous neuromorphic electronic devices have emerged in the last decade, most of them are rigid or contain materials that are toxic to biological systems. In this work, we report on biocompatible bilayer graphene-based artificial synaptic transistors (BLAST) capable of mimicking synaptic behavior. The BLAST devices leverage a dry ion-selective membrane, enabling long-term potentiation, with ~50 aJ/µm 2 switching energy efficiency, at least an order of magnitude lower than previous reports on two-dimensional material-based artificial synapses. The devices show unique metaplasticity, a useful feature for generalizable deep neural networks, and we demonstrate that metaplastic BLASTs outperform ideal linear synapses in classic image classification tasks. With switching energy well below the 1 fJ energy estimated per biological synapse, the proposed devices are powerful candidates for bio-interfaced online learning, bridging the gap between artificial and biological neural networks.

97 MATHEMATICS AND COMPUTING↗

Rapid and flexible segmentation of electron microscopy data using few-shot machine learning

Abstract Automatic segmentation of key microstructural features in atomic-scale electron microscope images is critical to improved understanding of structure–property relationships in many important materials and chemical systems. However, the present paradigm involves time-intensive manual analysis that is inherently biased, error-prone, and unable to accommodate the large volumes of data produced by modern instrumentation. While more automated approaches have been proposed, many are not robust to a high variety of data, and do not generalize well to diverse microstructural features and material systems. Here, we present a flexible, semi-supervised few-shot machine learning approach for segmentation of scanning transmission electron microscopy images of three oxide material systems: (1) epitaxial heterostructures of SrTiO 3 /Ge, (2) La 0.8 Sr 0.2 FeO 3 thin films, and (3) MoO 3 nanoparticles. We demonstrate that the few-shot learning method is more robust against noise, more reconfigurable, and requires less data than conventional image analysis methods. This approach can enable rapid image classification and microstructural feature mapping needed for emerging high-throughput characterization and autonomous microscope platforms.

36 MATERIALS SCIENCE↗

Topological structure of complex predictions

Abstract Current complex prediction models are the result of fitting deep neural networks, graph convolutional networks or transducers to a set of training data. A key challenge with these models is that they are highly parameterized, which makes describing and interpreting the prediction strategies difficult. We use topological data analysis to transform these complex prediction models into a simplified topological view of the prediction landscape. The result is a map of the predictions that enables inspection of the model results with more specificity than dimensionality-reduction methods such as tSNE and UMAP. The methods scale up to large datasets across different domains. We present a case study of a transformer-based model previously designed to predict expression levels of a piece of DNA in thousands of genomic tracks. When the model is used to study mutations in the BRCA1 gene, our topological analysis shows that it is sensitive to the location of a mutation and the exon structure of BRCA1 in ways that cannot be found with tools based on dimensionality reduction. Moreover, the topological framework offers multiple ways to inspect results, including an error estimate that is more accurate than model uncertainty. Further studies show how these ideas produce useful results in graph-based learning and image classification.

Computer Science↗

A domain wall-magnetic tunnel junction artificial synapse with notched geometry for accurate and efficient training of deep neural networks

Inspired by the parallelism and efficiency of the brain, several candidates for artificial synapse devices have been developed for neuromorphic computing, yet a nonlinear and asymmetric synaptic response curve precludes their use for backpropagation, the foundation of modern supervised learning. Spintronic devices—which benefit from high endurance, low power consumption, low latency, and CMOS compatibility—are a promising technology for memory, and domain-wall magnetic tunnel junction (DW-MTJ) devices have been shown to implement synaptic functions such as long-term potentiation and spike-timing dependent plasticity. In this work, we propose a notched DW-MTJ synapse as a candidate for supervised learning. Using micromagnetic simulations at room temperature, we show that notched synapses ensure the non-volatility of the synaptic weight and allow for highly linear, symmetric, and reproducible weight updates using either spin transfer torque (STT) or spin–orbit torque (SOT) mechanisms of DW propagation. We use lookup tables constructed from micromagnetics simulations to model the training of neural networks built with DW-MTJ synapses on both the MNIST and Fashion-MNIST image classification tasks. Accounting for thermal noise and realistic process variations, the DW-MTJ devices achieve classification accuracy close to ideal floating-point updates using both STT and SOT devices at room temperature and at 400 K. Our work establishes the basis for a magnetic artificial synapse that can eventually lead to hardware neural networks with fully spintronic matrix operations implementing machine learning.

42 ENGINEERING↗

Multi-scale structural analysis of proteins by deep semantic segmentation

Abstract Motivation Recent advances in computational methods have facilitated large-scale sampling of protein structures, leading to breakthroughs in protein structural prediction and enabling de novo protein design. Establishing methods to identify candidate structures that can lead to native folds or designable structures remains a challenge, since few existing metrics capture high-level structural features such as architectures, folds and conformity to conserved structural motifs. Convolutional Neural Networks (CNNs) have been successfully used in semantic segmentation—a subfield of image classification in which a class label is predicted for every pixel. Here, we apply semantic segmentation to protein structures as a novel strategy for fold identification and structure quality assessment. Results We train a CNN that assigns each residue in a multi-domain protein to one of 38 architecture classes designated by the CATH database. Our model achieves a high per-residue accuracy of 90.8% on the test set (95.0% average per-class accuracy; 87.8% average per-structure accuracy). We demonstrate that individual class probabilities can be used as a metric that indicates the degree to which a randomly generated structure assumes a specific fold, as well as a metric that highlights non-conformative regions of a protein belonging to a known class. These capabilities yield a powerful tool for guiding structural sampling for both structural prediction and design. Availability and implementation The trained classifier network, parser network, and entropy calculation scripts are available for download at https://git.io/fp6bd, with detailed usage instructions provided at the download page. A step-by-step tutorial for setup is provided at https://goo.gl/e8GB2S. All Rosetta commands, RosettaRemodel blueprints, and predictions for all datasets used in the study are available in the Supplementary Information. Supplementary information Supplementary data are available at Bioinformatics online.

Eguchi, Raphael R.↗

CMOS-Based Single-Cycle in-Memory XOR/XNOR

Big data applications are on the rise, and so is the number of data centers. The ever-increasing massive data pool needs to be periodically backed up in a secure environment. Moreover, a massive amount of securely backed-up data is required for training binary convolutional neural networks for image classification. XOR and XNOR operations are essential for large-scale data copy verification, encryption, and classification algorithms. The disproportionate speed of existing compute and memory units makes the von Neumann architecture inefficient to perform these Boolean operations. Compute-in-memory (CiM) has proved to be an optimum approach for such bulk computations. The existing CiM-based XOR/XNOR techniques either require multiple cycles for computing or add to the complexity of the fabrication process. Here, we propose a CMOS-based hardware topology for single-cycle in-memory XOR/XNOR operations. Our design provides at least 2× improvement in the latency compared with other existing CMOS-compatible solutions. We verify the proposed system through circuit/system-level simulations and evaluate its robustness using a 5000-point Monte Carlo variation analysis. This all-CMOS design paves the way for practical implementation of CiM XOR/XNOR at scaled technology nodes.

97 MATHEMATICS AND COMPUTING↗

Understanding and Estimating Error Propagation in Neural Networks for Scientific Data Analysis

Neural networks are increasingly integrated into scientific discovery, where input data reduction and model quantization play a key role in accelerating inference. However, understanding and mitigating the impact of these techniques on output error is critical for ensuring reliable results, particularly in tasks demanding high numerical precision. This paper introduces a comprehensive framework for optimizing neural network inference in scientific computing by combining data reduction and weight quantization while maintaining error-controlled outcomes. We develop theoretical analyses to bound error propagation under these reductions and propose a framework that balances computational performance with error constraints. Evaluation on real-world learning-based combustion simulations and satellite image classification demonstrates that our derived error bounds accurately predict observed errors while enabling significant computational speedup under our framework. This work highlights the potential for further leveraging advancements in modern lossy compression algorithms and hardware accelerators that support lower-precision formats.

He, Weiming [New Jersey Institute of Technology]↗