Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “LABELLING”

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 775 records · Page 43

The oxidative pentose phosphate pathway in photosynthesis: a tale of two shunts

CO 2 release in the light (R L ) and its presumed source, oxidative pentose phosphate pathways, were found to be insensitive to CO 2 concentration. The oxidative pentose phosphate pathways form glucose 6-phosphate (G6P) shunts that bypass the nonoxidative pentose phosphate reactions of the Calvin–Benson cycle. Using adenosine diphosphate glucose and uridine diphosphate glucose as proxies for labeling of G6P in the stroma and cytosol respectively, it was found that only the cytosolic shunt was active. Uridine diphosphate glucose, a proxy for cytosolic G6P, and 6-phosphogluconate (6PG) were significantly less labeled than Calvin–Benson cycle intermediates in the light. But ADP glucose, a proxy for stromal G6P, is labeled to the same degree as Calvin–Benson cycle intermediates and much greater than 6PG. A metabolically inert pool of sedoheptulose bisphosphate can slowly equilibrate keeping the label in sedoheptulose lower than in other stromal metabolites. Finally, phosphorylation of fructose 6-phosphate (F6P) in the cytosol can allow some unlabeled carbon in cytosolic F6P to dilute label in phosphenolpyruvate. The results clearly show that there is oxidative pentose phosphate pathway activity in the cytosol that provides a shunt around the nonoxidative pentose phosphate pathway reactions of the Calvin–Benson cycle and is not strongly CO 2 -sensitive.

59 BASIC BIOLOGICAL SCIENCES↗

TH-CD-206-02: BEST IN PHYSICS (IMAGING): 3D Prostate Segmentation in MR Images Using Patch-Based Anatomical Signature

Purpose: MRI has shown promise in identifying prostate tumors with high sensitivity and specificity for the detection of prostate cancer. Accurate segmentation of the prostate plays a key role various tasks: to accurately localize prostate boundaries for biopsy needle placement and radiotherapy, to initialize multi-modal registration algorithms or to obtain the region of interest for computer-aided detection of prostate cancer. However, manual segmentation during biopsy or radiation therapy can be time consuming and subject to inter- and intra-observer variation. This study’s purpose it to develop an automated method to address this technical challenge. Methods: We present an automated multi-atlas segmentation for MR prostate segmentation using patch-based label fusion. After an initial preprocessing for all images, all the atlases are non-rigidly registered to a target image. And then, the resulting transformation is used to propagate the anatomical structure labels of the atlas into the space of the target image. The top L similar atlases are further chosen by measuring intensity and structure difference in the region of interest around prostate. Finally, using voxel weighting based on patch-based anatomical signature, the label that the majority of all warped labels predict for each voxel is used for the final segmentation of the target image. Results: This segmentation technique was validated with a clinical study of 13 patients. The accuracy of our approach was assessed using the manual segmentation (gold standard). The mean volume Dice Overlap Coefficient was 89.5±2.9% between our and manual segmentation, which indicate that the automatic segmentation method works well and could be used for 3D MRI-guided prostate intervention. Conclusion: We have developed a new prostate segmentation approach based on the optimal feature learning label fusion framework, demonstrated its clinical feasibility, and validated its accuracy. This segmentation technique could be a useful tool in image-guided interventions for prostate-cancer diagnosis and treatment.

60 APPLIED LIFE SCIENCES↗

Self-Supervised Cloud Classification

Abstract Low-level marine clouds play a pivotal role in Earth’s weather and climate through their interactions with radiation, heat and moisture transport, and the hydrological cycle. These interactions depend on a range of dynamical and microphysical processes that result in a broad diversity of cloud types and spatial structures, and a comprehensive understanding of cloud morphology is critical for continued improvement of our atmospheric modeling and prediction capabilities moving forward. Deep learning has recently accelerated our ability to study clouds using satellite remote sensing, and machine learning classifiers have enabled detailed studies of cloud morphology. A major limitation of deep learning approaches to this problem, however, is the large number of hand-labeled samples that are required for training. This work applies a recently developed self-supervised learning scheme to train a deep convolutional neural network (CNN) to map marine cloud imagery to vector embeddings that capture information about mesoscale cloud morphology and can be used for satellite image classification. The model is evaluated against existing cloud classification datasets and several use cases are demonstrated, including training cloud classifiers with very few labeled samples, interrogation of the CNN’s learned internal feature representations, cross-instrument application, and resilience against sensor calibration drift and changing scene brightness. The self-supervised approach learns meaningful internal representations of cloud structures and achieves comparable classification accuracy to supervised deep learning methods without the expense of creating large hand-annotated training datasets. Significance Statement Marine clouds heavily influence Earth’s weather and climate, and improved understanding of marine clouds is required to improve our atmospheric modeling capabilities and physical understanding of the atmosphere. Recently, deep learning has emerged as a powerful research tool that can be used to identify and study specific marine cloud types in the vast number of images collected by Earth-observing satellites. While powerful, these approaches require hand-labeling of training data, which is prohibitively time intensive. This study evaluates a recently developed self-supervised deep learning method that does not require human-labeled training data for processing images of clouds. We show that the trained algorithm performs competitively with algorithms trained on hand-labeled data for image classification tasks. We also discuss potential downstream uses and demonstrate some exciting features of the approach including application to multiple satellite instruments, resilience against changing image brightness, and its learned internal representations of cloud types. The self-supervised technique removes one of the major hurdles for applying deep learning to very large atmospheric datasets.

54 ENVIRONMENTAL SCIENCES↗

Soil and soil fraction radiocarbon, C and N from the Enriched Background Isotope Study, Oak Ridge National Laboratory, Tennessee, USA

A large accidental 14CO2 release at the Oak Ridge Reservation in Tennessee, USA provided a strong label pulse into adjacent, well-studied oak forests, resulting in highly elevated Δ14C values in leaf litter (~1000‰) and roots (~260–450‰). A four-year manipulative study was conducted to provide insight into mechanisms underlying soil Organic matter dynamics, specifically to determine the relative contribution of litter versus roots to the bulk mineral soil C pool, as well as to free light, occluded light and heavy fractions. The heavy fraction was further split into fractions with densities of 1.7–2.4 g cm-3 and >2.4 g cm-3 to test the homogeneity of the mineral-associated fraction of C. Substantial concentrations of label were detected in all soil fractions within a year of the 14CO2 release, indicating rapid incorporation of newly fixed photosynthates in all fractions of soil organic C, regardless of differences in stability inferred by previous work. This rapid incorporation of label occurred only in treatments where roots were labeled, indicating that roots are the major source of inputs to mineral soil C stocks at these sites. Separation of the heavy fraction into subfractions of intermediate (1.7–2.4 g cm -3) and high (>2.4 g cm-3) density indicated that both subfractions incorporated label at similar rates, despite significant differences in degree of microbial processing.

54 ENVIRONMENTAL SCIENCES↗

Identifying Vehicle Signals in Continuous Seismic Data Using Unsupervised Machine-Learning Techniques

Seismic sensors deployed near roadways effectively capture ground vibrations generated by passing vehicles. Although both traditional and machine‐learning algorithms have been utilized for analyzing such signals, independent validation of detected vehicle events remains limited. We applied two unsupervised machine‐learning algorithms, uniform manifold approximation and projection for dimension reduction, and hierarchical density‐based spatial clustering of applications with noise, to continuous seismic data collected along a road on the main campus of Oak Ridge National Laboratory. The algorithms identified seven distinct cluster labels across the entire dataset. By comparing these cluster labels with precipitation records from a nearby weather station and image‐derived labels from a local camera system, we identified one cluster associated with rainfall and another with vehicle activity. Our algorithms identified a greater number of vehicle‐related labels compared to the camera‐derived labels because seismic data are unaffected by poor lighting conditions. The arrival times of the newly detected vehicle signals corresponded well with the road’s speed limit, supporting our findings. Our algorithm outperformed the short‐term average/long‐term average method and k‐means clustering. Our results suggest that seismic data, when analyzed with machine‐learning algorithms, can complement existing vehicle monitoring systems, particularly under challenging environmental conditions.

Chai, Chengping [Oak Ridge National Laboratory (OR↗

Expectation-maximization algorithms for learning a finite mixture of univariate survival time distributions from partially specified class values

Heterogeneity exists on a data set when samples from di erent classes are merged into the data set. Finite mixture models can be used to represent a survival time distribution on heterogeneous patient group by the proportions of each class and by the survival time distribution within each class as well. The heterogeneous data set cannot be explicitly decomposed to homogeneous subgroups unless all the samples are precisely labeled by their origin classes; such impossibility of decomposition is a barrier to overcome for estimating nite mixture models. The expectation-maximization (EM) algorithm has been used to obtain maximum likelihood estimates of nite mixture models by soft-decomposition of heterogeneous samples without labels for a subset or the entire set of data. In medical surveillance databases we can find partially labeled data, that is, while not completely unlabeled there is only imprecise information about class values. In this study we propose new EM algorithms that take advantages of using such partial labels, and thus incorporate more information than traditional EM algorithms. We particularly propose four variants of the EM algorithm named EM-OCML, EM-PCML, EM-HCML and EM-CPCML, each of which assumes a specific mechanism of missing class values. We conducted a simulation study on exponential survival trees with five classes and showed that the advantages of incorporating substantial amount of partially labeled data can be highly signi cant. We also showed model selection based on AIC values fairly works to select the best proposed algorithm on each specific data set. A case study on a real-world data set of gastric cancer provided by Surveillance, Epidemiology and End Results (SEER) program showed a superiority of EM-CPCML to not only the other proposed EM algorithms but also conventional supervised, unsupervised and semi-supervised learning algorithms.

97 MATHEMATICS AND COMPUTING↗

The Hunt for Green Every April: Factors Affecting Fitness in Switchgrass

This grant funded work was undertaken to develop fundamental biological knowledge of the factors affecting the complex plant trait “fitness” in switchgrass (Panicum virgatum L.), a plant being developed as a biomass crop. Using a diverse range of latitudinally-adapted switchgrass plants, genomic, molecular and physiological studies were performed to track a number of different aspects of plant genetics and physiology over the course of the growing season. Work was performed on both genetically unrelated and genetically related plants. Plants were established in the field from seedlings raised in a greenhouse, or from clones present in other field nurseries. Field grown plants were used as the source of all tissues. The three objectives of this proposal were:(1) Transcript Profiling, Metabolomics, and C and N Partitioning and Recycling in Crowns and Rhizomes of Switchgrass over two growing seasons; (2) Gene Profiling During Regreening and Dormancy of Bulked Segregants; (3) Extent of Linkage Disequilibrium in Populations for Adaptation and Fitness Traits Being Developed for Central and Northern USA, that Show Significant Heterosis. Objective 1 results: Plants were labeled using 13 CO2 (a stable isotope) using an acrylic chamber constructed specifically for this purpose. Plants became labeled with 13 C and label decayed in aerial tissues over the course of the growing season. Varying amounts of 13 C were recovered in the rhizomes. These data are being analyzed. Plants were also labeled with 15 N-urea. Plants absorbed significant amounts of label that was remobilized to the growing shoots. N-dynamics would suggest that a portion of the 15 N absorbed into the crowns and rhizomes is sequestered below ground. Variable amounts of 15 N were translocated from the shoots to the roots over the course of the growing season. Polar metabolites extracted from a diverse array of rhizomes were analyzed using GCMS. Data indicated that there was a significant shift in metabolite pools over the course of the growing season, and differences in the levels of specific metabolites could be linked to the progression of dormancy. Several metabolites that accumulate in dormant rhizomes were identified. Some of these metabolites could be potentially linked to winter-survival of switchgrass. Extensive high-throughput sequencing was conducted on crown and rhizome samples collected from field grown plants. Initial work was performed on a Roche 454 system. All later work was performed on an Illumina sequencing-by-synthesis system. Some of these datasets have been published as peer-reviewed papers, other data are currently being analyzed and being readied for publication. Objective 2 results: Genetically related but phenotypically divergent plants from an octaploid switchgrass population were grown in a replicated field nursery. Rhizomes were harvested at four different times over the course of the growing season from plants with high winter survival and those with lower winter survival. RNA-Seq was performed on harvested materials. Initial analysis suggests that plants with lowered winter survival experience a greater level of cellular stress in dormant tissues. This aspect of plant function is being probed in greater depth. Objective 3 results: A total of 592 individual clones with three clonal replications in a randomized complete block design from each of five populations used in Objective 1 studies were rated for heading date in 2012 and 2014, green-up day of year in 2013, anthesis date in 2012, and yield in 2012, they were also subjected to NIR spectroscopy to derive cell wall composition estimates based on prior NIR calibrations. Plants were genotyped via a genotyping by sequencing (GBS) approach from reduced representation libraries constructed with adaptors that identified each individual. Libraries generated with the restriction enzyme PstI and called SNPs using Samtools after alignment to version 1.1 of the switchgrass genome sequence. A total of approximately 40,000 SNPs were found. These were then further filtered to eliminate markers with a minor allele frequency of < 0.05. The results of population analysis using STRUCTURE with expected population sizes or cluster numbers (K), clearly shows the hybrid composition of the KxS population and discriminated easily between upland (Summer) and lowland (Kanlow) populations. Under an assumption of 5 distinct populations there were detectable differences in allele frequencies between subpopulations within the three Kanlow populations particularly with respect to Kanlow EM and Kanlow base. We detected 110 SNPs with an allele frequency difference of ≥ 0.2 between Kanlow EM and Kanlow base populations, while 120 SNPs showed an allele frequency difference of ≥ 0.15 between Kanlow N1 and Kanlow base populations. These data are being readied for publication.

59 BASIC BIOLOGICAL SCIENCES↗

Chloroplast Dynamics and Photosynthetic Efficiency: Final Technical Report

This project investigated the mechanism by which chloroplasts position themselves to maximize solar energy utilization, to enhance gas exchange, to minimize environmental stress, and to promote efficient exchange of metabolites with other compartments within the plant cell. Chloroplasts move within leaf cells to optimize light levels, moving toward levels of light useful for photosynthesis while moving away from excess light. Plastids sometimes extend their reach by sending out projections (stromules) that can connect anchor chloroplasts in position within the cell or provide close contacts with plasma membrane, mitochondria, peroxisomes, endoplasmic reticulum, and the nucleus. The intracellular location of chloroplasts in relation to other organelles with which they share biosynthetic pathways, such as peroxisomes and mitochondria in photorespiration, affects metabolite flow. This work contributed to the knowledge of the mechanisms of organelle movement and anchoring in specific locations in plant cells and how proteins traffic within the cell. We identified two domains on 12 of the 13 Arabidopsis myosins that were similar to the vacuole-binding (V) domain characterized in yeast and to the DIL domain characterized in yeast and mouse as required for secretory vesicle or melanosome movement, respectively. Because all of the Arabidopsis regions with homology to the V domain contain the amino acid sequence PAL, we refer to this region as the Arabidopsis PAL domain. We have used the yeast Myo2p tail structural information to model the 12 myosin XI tail domains containing the homologous PAL and DIL domains. Eight YFP::DIL domain fusions labeled peroxisomes; none labeled mitochondria or chloroplasts. Six myosin XI Vacuole domains labeled mitochondria and seven labeled Golgi bodies. The Arabidopsis myosin XI-F PAL domain and the homologous myosin XI-F PAL domain from N. benthamiana labels chloroplasts and stromules in N. benthamiana leaves. Using an Arabidopsis line containing hotoconvertible GFP, we observed transfer of protein from one plastid to another and within a stromule from single plastids. We provided time-lapse movies demonstrating movement of both the photoconvertible GFP and standard GFP between plastids. We previously demonstrated the lack of a plastid network within plant cells. We provided protocols explaining how to use fluorescent protein technology to track plastids and stromules within plant cells. We demonstrated that standard GFP unexpectedly could be photoconverted to a red form under certain conditions, allowing the use of GFP lines for studies that require photoconversion.

59 BASIC BIOLOGICAL SCIENCES↗

Particle integrity, sampling, and application of a DNA-tagged tracer for aerosol transport studies

Aerosols are an ever-present part of our daily environment and have extensive effects on both human and environmental health. Particles in the inhalable range (1-10 μm diameter) are of particular concern because their deposition in the lung can lead to a variety of illnesses including allergic reactions, viral or bacterial infections, and cancer. Understanding the transport of inhalable aerosols across both short and long distances is necessary to predict human exposures to aerosols. To assess the transport of hazardous aerosols, surrogate tracer particles are required to measure their transport through occupied spaces. These tracer particles must not only possess similar transport characteristics to those of interest but also be easily distinguished from the background at low levels and survive the environmental conditions of the testing environment. A previously-developed DNA-tagged particle (DNATrax), composed of food-grade sugar and a DNA oligonucleotide as a “barcode” label, shows promise as a new aerosol tracer. Herein, the use of DNATrax material is validated for use in both indoor and outdoor environments. Utilizing passive samplers made of materials commonly found in indoor environments followed by quantitative polymerase chain reaction (qPCR) assay for endpoint particle detection, particles detection was achieved up to 90 m from the aerosolization location and across shorter distances with high spatial resolution. The unique DNA label and PCR assay specificity were leveraged to perform multiple simultaneous experiments. This allowed the assessment of experimental reproducibility, a rare occurrence among aerosol field tests. To transition to outdoor testing, the solid material provides some protection of the DNA label when exposed to ultraviolet (UV) radiation, with 60% of the DNA remaining intact after 60 minutes under a germicidal lamp and the rate of degradation declining with irradiation time. Additionally, exposure of the DNATrax material using formulations of two different food-grade sugars (maltodextrin and erythritol) to humidity as high as 66% had no significant effect on the DNA label’s degradation or the particle’s aerodynamic diameter, confirming particle stability under such conditions. In summary, confirmation of the DNATrax particles’ size and label integrity under variable conditions combined with experiment multiplexing and high resolution sampling provides a powerful experimental design for modeling aerosol transport through occupied indoor and outdoor locations.

54 ENVIRONMENTAL SCIENCES↗

FY24 Progress Report: SRNL Analysis of ICCWR LCM and WAMS data for Corrosion and Cracking

Algorithms for Machine Learning (ML) and data analysis for the 3013 Surveillance Program have been developed in an ongoing collaborative effort by the Savannah River National Laboratory (SRNL) and the University of South Carolina (USC). The objective of the algorithms is to automate the identification of corrosion and crack formation in the Inner Container Closure Weld Region (ICCWR) of the canister system used to store Pu-bearing material. Data for corrosion and cracking is collected from large binary files generated by a Laser Confocal Microscope (LCM), the Wide Area 3D Measurement System (WAMS), or,in a recent proposal, by a Scanning Electron Microscope (SEM). The ML software uses the physical attributes in the data files (e.g., one or more of: height, color, and 16-bit grayscale values as functions of position in a plane projection) to detect signs of surface corrosion and cracking after being trained on similar data, with the features to be detected. Although the initial scope included screening for broader indicators of corrosion, e.g., pitting, identification of potential cracks was prioritized for the past several years at the request of program leadership. Labeled training data is essential to developing the ML algorithm, and enhancements to data labeling capability have been developed to address this essential precursor to application of ML routines. Efficient labeling is particularly important in view of the large volume of data required to train ML algorithms and the relative rarity of cracks in the ICCWR data set. The updated program will read binary data from either LCM, WAMS or SEM files, interrogate data attributes, facilitate user labeling of data for training ML algorithms, execute ML algorithms, output parameters from trained ML algorithms, report ML model accuracy with respect to labeled data, and generate graphical representations for various analyses. In FY24, hourglass neural networks (HNNs) that were initiated in FY22 were further developed and tested using available LCM data, and their performance was tested against that of the alternative U-Net Neural Network algorithm structure. HNNs along with previously developed Convolutional Neural Networks (CNNs) and Deep Neural Networks (DNNs) comprise a suite of ML tools for identification of cracks in the ICCWR

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Evaluation of Candidate Theranostics for 227 Th/ 89 Zr Paired Radioimmunotherapy of Lymphoma

227 Th is a promising radioisotope for targeted α-particle therapy. It produces 5 α-particles through its decay, with the clinically approved 223 Ra as its first daughter. There is an ample supply of 227 Th, allowing for clinical use; however, the chemical challenges of chelating this large tetravalent f-block cation are considerable. Using the CD20-targeting antibody ofatumumab, we evaluated chelation of 227 Th 4+ for α-particle–emitting and radiotheranostic applications. Methods: We compared 4 bifunctional chelators for thorium radiopharmaceutical preparation: S-2-(4-Isothiocyanatobenzyl)-1,4,7,10-tetraazacyclododecane tetraacetic acid (p-SCN-Bn-DOTA), 2-(4-isothicyanatobenzyl)-1,2,7,10,13-hexaazacyclooctadecane-1,4,7,10,13,16-hexaacetic acid (p-SCN-Bn-HEHA), p-isothiacyanatophenyl-1-hydroxy-2-oxopiperidine-desferrioxamine (DFOcyclo*-p-Phe-NCS), and macrocyclic 1,2-HOPO N-hydroxysuccinimide (L804-NHS). Immunoconstructs were evaluated for yield, purity, and stability in vitro and in vivo. Tumor targeting of the lead 227 Th-labeled compound in vivo was performed in CD20-expressing models and compared with a companion 89 Zr-labeled PET agent. Results: 227 Th-labeled ofatumumab-chelator constructs were synthesized to a radiochemical purity of more than 95%, excepting HEHA. 227 Th-HEHA-ofatumumab showed moderate in vitro stability. 227 Th-DFOcyclo*-ofatumumab presented excellent 227 Th labeling efficiency; however, high liver and spleen uptake was revealed in vivo, indicative of aggregation. 227 Th-DOTA-ofatumumab labeled poorly, yielding no more than 5%, with low specific activity (0.08 GBq/g) and modest long-term in vitro stability (<80%). 227 Th-L804-ofatumumab coordinated 227 Th rapidly and efficiently at high yields, purity, and specific activity (8 GBq/g) and demonstrated extended stability. In vivo tumor targeting confirmed the utility of this chelator, and the diagnostic analog, 89 Zr-L804-ofatumumab, showed organ distribution matching that of 227 Th to delineate SU-DHL-6 tumors. Conclusion: Commercially available and novel chelators for 227 Th showed a range of performances. The L804 chelator can be used with potent radiotheranostic capabilities for 89 Zr/ 227 Th quantitative imaging and α-particle therapy.

227Th↗

SNM Radiation Signature Classification Using Different Semi-Supervised Machine Learning Models

The timely detection of special nuclear material (SNM) transfers between nuclear facilities is an important monitoring objective in nuclear nonproliferation. Persistent monitoring enabled by successful detection and characterization of radiological material movements could greatly enhance the nuclear nonproliferation mission in a range of applications. Supervised machine learning can be used to signal detections when material is present if a model is trained on sufficient volumes of labeled measurements. However, the nuclear monitoring data needed to train robust machine learning models can be costly to label since radiation spectra may require strict scrutiny for characterization. Therefore, this work investigates the application of semi-supervised learning to utilize both labeled and unlabeled data. As a demonstration experiment, radiation measurements from sodium iodide (NaI) detectors are provided by the Multi-Informatics for Nuclear Operating Scenarios (MINOS) venture at Oak Ridge National Laboratory (ORNL) as sample data. Anomalous measurements are identified using a method of statistical hypothesis testing. After background estimation, an energy-dependent spectroscopic analysis is used to characterize an anomaly based on its radiation signatures. In the absence of ground-truth information, a labeling heuristic provides data necessary for training and testing machine learning models. Supervised logistic regression serves as a baseline to compare three semi-supervised machine learning models: co-training, label propagation, and a convolutional neural network (CNN). In each case, the semi-supervised models outperform logistic regression, suggesting that unlabeled data can be valuable when training and demonstrating value in semi-supervised nonproliferation implementations.

98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL P↗

Contrastive Machine Learning with Gamma Spectroscopy Data Augmentations for Detecting Shielded Radiological Material Transfers

Data analysis techniques can be powerful tools for rapidly analyzing data and extracting information that can be used in a latent space for categorizing observations between classes of data. Machine learning models that exploit learned data relationships can address a variety of nuclear nonproliferation challenges like the detection and tracking of shielded radiological material transfers. The high resource cost of manually labeling radiation spectra is a hindrance to the rapid analysis of data collected from persistent monitoring and to the adoption of supervised machine learning methods that require large volumes of curated training data. Instead, contrastive self-supervised learning on unlabeled spectra can enhance models that are built on limited labeled radiation datasets. This work demonstrates that contrastive machine learning is an effective technique for leveraging unlabeled data in detecting and characterizing nuclear material transfers demonstrated on radiation measurements collected at an Oak Ridge National Laboratory testbed, where sodium iodide detectors measure gamma radiation emitted by material transfers between the High Flux Isotope Reactor and the Radiochemical Engineering Development Center. Label-invariant data augmentations tailored for gamma radiation detection physics are used on unlabeled spectra to contrastively train an encoder, learning a complex, embedded state space with self-supervision. A linear classifier is then trained on a limited set of labeled data to distinguish transfer spectra between byproducts and tracked nuclear material using representations from the contrastively trained encoder. The optimized hyperparameter model achieves a balanced accuracy score of 80.30%. Any given model—that is, a trained encoder and classifier—shows preferential treatment for specific subclasses of transfer types. Regardless of the classifier complexity, a supervised classifier using contrastively trained representations achieves higher accuracy than using spectra when trained and tested on limited labeled data.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

High-Dimensional Bayesian Optimization via Semi-Supervised Learning with Optimized Unlabeled Data Sampling

We introduce a novel semi-supervised learning approach, named Teacher-Student Bayesian Optimization (TSBO ), integrating the teacher-student paradigm into BO to minimize expensive labeled data queries for the first time. TSBO incorporates a teacher model, an unlabeled data sampler, and a student model. The student is trained on unlabeled data locations generated by the sampler, with pseudo labels predicted by the teacher. The interplay between these three components implements a unique selective regularization to the teacher in the form of student feedback. This scheme enables the teacher to predict high-quality pseudo labels, enhancing the generalization of the GP surrogate model in the search space. To fully exploit TSBO , we propose two optimized unlabeled data samplers to construct effective student feedback that well aligns with the objective of Bayesian optimization. Furthermore, we quantify and leverage the uncertainty of the teacher-student model for the provision of reliable feedback to the teacher in the presence of risky pseudo-label predictions. TSBO demonstrates significantly improved sample-efficiency in several global optimization tasks under tight labeled data budgets. The implementation is available at https://github.com/reminiscenty/TSBO-Official.

Yin, Yuxuan↗

Contrasting colonization and plant growth promoting capacity between wild type and gfp-derative of the endophyte Pseudomonas putida W619 in hybrid poplar

This study aims to investigate the colonization of poplar by the endophyte Pseudomonas putida W619 and its capacity to promote plant growth. Poplar cuttings were inoculated with P. putida W619 (wild-type or gfp-labelled). The colonization of both strains was investigated and morphological, physiological and biochemical parameters were analyzed to evaluate plant growth promotion. Inoculation with P. putida W619 (wild-type) resulted in remarkable growth promotion, decreased activities of antioxidative defence related enzymes, and reduced stomatal resistance, all indicative of improved plant health and growth in comparison with the non-inoculated cuttings. In contrast, inoculation with gfp-labelled P. putida W619 did not promote growth; it even had a negative effect on plant health and growth. Furthermore, compared to the wildtype strain, colonization by the gfp-labelled P. putida W619::gfp1 was much lower; it only colonized the rhizosphere and root cortex while the wild-type strain also colonized the root xylem vessels. Despite the strong plant growth promoting capacity of P. putida W619 (wild-type), after gfp labelling its growth promoting characteristics disappeared and its colonization capacity was strongly influenced; for these reasons gfp labelling should be applied with sufficient caution.

09 BIOMASS FUELS↗

Combinatorial Information Theoretical Measurement of the Semantic Significance of Semantic Graph Motifs

Given an arbitrary semantic graph data set, perhaps one lacking in explicit ontological information, we wish to first identify its significant semantic structures, and then measure the extent of their significance. Casting a semantic graph dataset as an edge-labeled, directed graph, this task can be built on the ability to mine frequent {\em labeled} subgraphs in edge-labeled, directed graphs. We begin by considering the fundamentals of the enumerative combinatorics of subgraph motif structures in edge-labeled directed graphs. We identify its frequent labeled, directed subgraph motif patterns, and measure the significance of the resulting motifs by the information gain relative to the expected value of the motif based on the empirical frequency distribution of the link types which compose them, assuming indpendence. We illustrate the method on a small test graph, and discuss results obtained for small linear motifs (link type bigrams and trigrams) in a larger graph structure.

Joslyn, Cliff A.↗

Classifying with confidence from incomplete information.

For this paper, we consider the problem of classifying a test sample given incomplete information. This problem arises naturally when data about a test sample is collected over time, or when costs must be incurred to compute the classification features. For example, in a distributed sensor network only a fraction of the sensors may have reported measurements at a certain time, and additional time, power, and bandwidth is needed to collect the complete data to classify. A practical goal is to assign a class label as soon as enough data is available to make a good decision. We formalize this goal through the notion of reliability—the probability that a label assigned given incomplete data would be the same as the label assigned given the complete data, and we propose a method to classify incomplete data only if some reliability threshold is met. Our approach models the complete data as a random variable whose distribution is dependent on the current incomplete data and the (complete) training data. The method differs from standard imputation strategies in that our focus is on determining the reliability of the classification decision, rather than just the class label. We show that the method provides useful reliability estimates of the correctness of the imputed class labels on a set of experiments on time-series data sets, where the goal is to classify the time-series as early as possible while still guaranteeing that the reliability threshold is met.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Generative memory for lifelong machine learning

Techniques are disclosed for training machine learning systems. An input device receives training data comprising pairs of training inputs and training labels. A generative memory assigns training inputs to each archetype task of a plurality of archetype tasks, each archetype task representative of a cluster of related tasks within a task space and assigns a skill to each archetype task. The generative memory generates, from each archetype task, auxiliary data comprising pairs of auxiliary inputs and auxiliary labels. A machine learning system trains a machine learning model to apply a skill assigned to an archetype task to training and auxiliary inputs assigned to the archetype task to obtain output labels corresponding to the training and auxiliary labels associated with the training and auxiliary inputs assigned to the archetype task to enable scalable learning to obtain labels for new tasks for which the machine learning model has not previously been trained.

Nadamuni Raghavan, Aswin↗