Search NASA⌕ Search

SEARCH · Search NASA

Results for “Self Organizing Map”

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 127 records · Page 7

Combining multiple models to generate consensus: Application to radiation-induced pneumonitis prediction

The fusion of predictions from disparate models has been used in several fields to obtain a more realistic and robust estimate of the ''ground truth'' by allowing the models to reinforce each other when consensus exists, or, conversely, negate each other when there is no consensus. Fusion has been shown to be most effective when the models have some complementary strengths arising from different approaches. In this work, we fuse the results from four common but methodologically different nonlinear multivariate models (Decision Trees, Neural Networks, Support Vector Machines, Self-Organizing Maps) that were trained to predict radiation-induced pneumonitis risk on a database of 219 lung cancer patients treated with radiotherapy (34 with Grade 2+ postradiotherapy pneumonitis). Each model independently incorporated a small number of features from the available set of dose and nondose patient variables to predict pneumonitis; no two models had all features in common. Fusion was achieved by simple averaging of the predictions for each patient from all four models. Since a model's prediction for a patient can be dependent on the patient training set used to build the model, the average of several different predictions from each model was used in the fusion (predictions were made by repeatedly testing each patient with a model built from different cross-validation training sets that excluded the patient being tested). The area under the receiver operating characteristics curve for the fused cross-validated results was 0.79, with lower variance than the individual component models. From the fusion, five features were extracted as the consensus among all four models in predicting radiation pneumonitis. Arranged in order of importance, the features are (1) chemotherapy; (2) equivalent uniform dose (EUD) for exponent a=1.2 to 3; (3) EUD for a=0.5 to 1.2, lung volume receiving >20-30 Gy; (4) female sex; and (5) squamous cell histology. To facilitate ease of interpretation and prospective use, the fused outcome results for the patients were fitted to a logistic probability function.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Deciphering the molecular components of the quorum sensing system in the fungus Ophiostoma piceae

Quorum sensing (QS) is a complex cell-cell communication mechanism that coordinates population-level behaviors in microbes. In eukaryotes, this phenomenon has been extensively described in the dimorphic yeast Candida albicans as its main QS molecule, the sesquiterpene alcohol farnesol, is responsible for various phenotypic (i.e., inhibition of yeast-to-hyphae transition, biofilm formation, and, hence, pathogenesis) and metabolic (i.e., induction of oxidative stress and apoptosis) changes. Ophiostoma piceae CECT 20416 is a dimorphic saprotrophic ascomycete with biotechnological interest that also produces farnesol as QS molecule, but in this case, the alcohol promotes the morphological transition to the mycelial form, biofilm formation, enzyme secretion, and melanin production. Here, we characterized the physiological response of Ophiostoma piceae to farnesol, the molecular components of the QS system of this fungus have been investigated using a “multiomics” approach that involved genomic, transcriptomic, and proteomic analyses. Some genes identified in this work are proposed as key factors in farnesol transport and signaling. We have also cataloged the genes undergoing major transcriptional changes triggered by the presence of the autoinducer, such as cell-wall remodeling, reactive oxygen species (ROS) protection, and melanin biosynthesis, using self-organizing maps. This analysis could be useful for applications in the forestry industry, for enzymes production, and for the valorization of residues. Furthermore, it might as well help to investigate the QS mechanisms of clinically relevant fungi phylogenetically related to Ophiostoma.

59 BASIC BIOLOGICAL SCIENCES↗

OpenCRUMS USA: An Open Machine Learning Framework for Characterizing Variability in Aerosol Reanalysis Data

Advances in artificial intelligence (AI) have called for exploring how these techniques can be used for exploring patterns in large climate datasets. To that regard, the U.S. Department of Energy AI for Earth System Predictability (AI4ESP) supported a pilot initiative called the Open Classification of Regimes in the Southeast USA (OpenCRUMS USA) project to explore how AI can be used to characterize modes of spatial variability in large climate datasets. For this study, we focus on comparing two methods for characterizing the modes of spatial variability of surface aerosol concentration over the Houston region: empirical orthogonal functions (EOFs) and layerwise relevance propagation (LRP) applied to a convolutional neural network (CNN) classifier. We show that EOF analysis typically attributes spatial variability modes that span all of southeast Texas, prohibiting the attribution of spatial variability to localized regions. However, using LRP on the CNN classifier resolves the explanatory parameters at a finer spatial resolution than EOFs. This allows for the attribution of the spatial variability of surface aerosols to local regions of organic carbon which was not possible using EOFs. In addition, the LRP analysis also suggests that synoptic-scale transport of dust is most prevalent during anticyclonic and pretrough synoptic conditions as categorized by self-organizing maps.

54 ENVIRONMENTAL SCIENCES↗

Joint Synoptic and Cloud Variability over the Northeast Atlantic near the Azores

Abstract Marine boundary layer clouds are modified by processes at different spatial and temporal scales. To isolate the processes governing aerosol–cloud–precipitation interactions, multiday synoptic variability of the environment must be accounted for. Information on the location of low clouds relative to the ridge–trough pattern gives insight into how cloud properties vary as a function of environmental subsidence and stability. The technique of self-organizing maps (SOMs) is employed to objectively classify the 500-hPa geopotential height patterns for 33 years of reanalysis fields (ERA-Interim) into pretrough, trough, posttrough, ridge, and zonal-flow categories. The SOM technique is applied to a region of prevalent marine low cloudiness over the eastern North Atlantic Ocean that is centered on the Azores island chain, the location of a long-term U.S. Department of Energy observation site. The Azores consistently lie in an area of substantial variability in synoptic configuration, thermodynamic environment, and cloud properties. The SOM method was run in two ways to emphasize multiday and seasonal variability separately. Over and near the Azores, there is an east-to-west sloshing back and forth of the western edge of marine low clouds associated with different synoptic states. The different synoptic states also exhibit substantial north–south variability in the position of high clouds. For any given month of the year, there is large year-to-year variability in the occurrence of different synoptic states. Hence, estimating the climatological behavior of clouds from short-term field campaigns has large uncertainties. This SOM approach is a robust method that is broadly applicable to characterizing synoptic regimes for any location.

54 ENVIRONMENTAL SCIENCES↗

Observed Temperature and Moisture Advection Characteristics in the Southern Great Plains

Until recently, advection profiles in the planetary boundary layer have been difficult to quantify from observations. As a result, not much is known about its basic characteristics, like magnitude and diurnal cycles, or how these differ as a function of large-scale environments. Here, to provide insight into advective characteristics, a Green’s Theorem–based method for calculating profiles of advection in the lowest 3 km from an array of ground-based thermodynamic and kinematic profiling instruments has been applied to 2 years of observations from the Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) observing site. Since advection largely depends on synoptic-scale forcing, a self-organizing map (SOM) was used to classify days based on mean sea level pressure analyses; from that, the diurnal evolution of vertical profiles of advection for different synoptic environments was quantified. The overall mean magnitude of the potential temperature advection is approximately ±3 K h −1 , while the moisture advection is approximately ±1.5 g kg −1 h −1 , with substantial variability in the sign and magnitude of the advective tendencies at different heights and throughout the diurnal cycle. Advection magnitude is strongly connected to the strength of synoptic-scale forcing, which varies depending on the time of year. Variability in advection is larger for potential temperature than it is for moisture, and strongly forced environments, like midlatitude cyclones, feature greater variability in the advection magnitude than weakly forced or quiescent environments.

advection↗

Contrasting spring and summer large-scale environments associated with mesoscale convective systems over the U.S. Great Plains

Mesoscale convective systems (MCSs) are frequently observed over the U.S. Great Plains during boreal spring and summer. Here, four types of synoptically-favorable environments for spring MCSs and two types of synoptically- favorable and unfavorable environments for summer MCSs are identified using self-organizing maps (SOM) with the inputs of observational data. During spring, frontal systems providing a lifting mechanism and an enhanced Great Plains low-level jet (GPLLJ) providing anomalous moisture are important features identified by SOM analysis as they create favorable dynamical and thermodynamic environments for MCS development. During summer, the composite MCS environment shows small positive convective available potential energy (CAPE) and convective inhibition (CIN) anomalies, which are in stark contrast with the large positive CAPE and negative CIN anomalies in spring. This contrast suggests that convection may occur even within weak dynamical and thermodynamic perturbations in summer so that MCSs may be inherently less predictable in summer. The two synoptically-favorable environments identified in summer have frontal characteristics and an enhanced GPLLJ, but both shift north compared to spring. The two synoptically-unfavorable environments feature enhanced upper-level ridges, but differ in the strength of the GPLLJ. In both seasons, MCS precipitation amount, area and rate are much larger in the frontal-related MCSs than non-frontal MCSs. A large-scale index constructed using the pattern correlation between the large-scale environment and the synoptically-favorable SOM types is found to be skillful for predicting MCS number, precipitation rate and area in spring, but its explanatory power decreases significantly in summer. The low predictability of summer MCSs deserves further investigation in the future.

Song, Fengfei↗

Northern Hemisphere Winter Air Temperature Patterns and Their Associated Atmospheric and Ocean Conditions

The Northern Hemisphere (NH) has experienced winter Arctic warming and continental cooling in recent decades, but the dominant patterns in winter surface air temperature (SAT) are not well understood. Here, a self-organizing map (SOM) analysis is performed to identify the leading patterns in winter daily SAT fields from 1979 to 2018, and their associated atmospheric and ocean conditions are also examined. Three distinct winter SAT patterns with two phases of nearly opposite signs and a time scale of 7–12 days are found: one pattern exhibits concurrent SAT anomalies of the same sign over North America (NA) and northern Eurasia, while the other two patterns show SAT anomalies of opposite signs between, respectively, NA and the Bering Sea, and the Kara Sea and East Asia (EA). Winter SAT variations may arise from changes in the SOM frequencies. Specifically, the observed increasing trends of winter cold extremes over NA, central Eurasia, and EA during 1998–2013 can be understood as a result of the increasing occurrences of some specific SAT patterns. These SOMs are closely related to poleward advection of midlatitude warm air and equatorward movements of polar cold airmass. These meridional displacements of cold and warm airmasses cause concurrent anomalies over different regions not only in SAT but also in water vapor and surface downward longwave radiation. Anomalous sea surface temperatures in the tropical Pacific, midlatitude North Pacific, and North Atlantic and anomalous Arctic sea ice concentrations also concur to support and maintain the anomalous atmospheric circulation that causes the SAT anomalies.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Seasonal and Geographical Variations in Fundamental Weather Patterns during Extreme Precipitation as Identified from Omega Equation Forcing

Abstract We investigate the large-scale weather patterns during extreme precipitation (PEx) events over the conterminous United States (CONUS) by applying a version of the quasigeostrophic (QG) omega equation. This work aims to develop a climatology of the weather patterns most related to PEx events during current climate. Extreme events are examined for each of seven regions defined by consistent annual cycles of precipitation and spanning the CONUS. For the CONUS we train several self-organizing maps (SOM) on a pressure–time series of vertical velocity from each of the advective forcing terms in the QG omega equation for each extreme event. The unsupervised learning of the SOM allows us to identify the most descriptive set of nine patterns in vertical velocity associated with precipitation extremes. This method finds multiple frontal- and cyclone-driven patterns while grouping primarily convective events into one pattern. Frontal events include a synoptic pattern consistent with West Coast atmospheric river events as well as pattern groups linked to developing and to mature (“occluded”) frontal cyclones. The primary patterns found during PEx events vary seasonally and geographically. Frontal cyclone patterns are most common during PEx events during summer in the part of the Great Plains and during winter for the Northeast, Southeast, Pacific Northwest, and Southwest. Convection is the most common pattern during summer in all regions. Except in the Southeast, the annual cycles of monthly number of PEx events and average precipitation match well, partially validating our choice of regions to aggregate PEx events.

Meteorology & Atmospheric Sciences↗

Clusters of Regional Precipitation Seasonality Change in the Community Earth System Model, Version 2

Abstract The likely changes to precipitation seasonality with warming are both impactful and not well understood. This work aims to describe areas that experience similar changes to seasonal precipitation irrespective of the original underlying precipitation seasonality. We train a self-organizing map on the difference between the seasonal cycle of precipitation in the past and in a high-warming future climate as represented by the Community Earth System Model, version 2, to create regions with similar changes in precipitation seasonality. This method is applied separately over land and ocean surfaces because of the differing processes leading to precipitation over each. This method indicates that future changes in seasonal precipitation are most varied in the tropics because of a southward shift in the intertropical convergence zone. The seasonal shifts found over midlatitude oceans indicate a poleward shift in atmospheric river activity. We find a correspondence between certain land-based precipitation changes and Köppen climate classification. The seasonality of large-scale and convective precipitation is examined for each region. The relationship between the seasonal changes to precipitation and associated atmospheric processes is discussed. These processes include atmospheric rivers, the intertropical convergence zone, tropical cyclones, and monsoons.

54 ENVIRONMENTAL SCIENCES↗

Patterns of gene expression associated with recovery and injury in heat-stressed rats

Background: The in vivo gene response associated with hyperthermia is poorly understood. Here, we perform a global, multiorgan characterization of the gene response to heat stress using an in vivo conscious rat model. Results: We heated rats until implanted thermal probes indicated a maximal core temperature of 41.8°C (T c,Max ). We then compared transcriptomic profiles of liver, lung, kidney, and heart tissues harvested from groups of experimental animals at T c,Max , 24 hours, and 48 hours after heat stress to time-matched controls kept at an ambient temperature. Cardiac histopathology at 48 hours supported persistent cardiac injury in three out of six animals. Microarray analysis identified 78 differentially expressed genes common to all four organs at T c,Max . Self-organizing maps identified gene-specific signatures corresponding to protein-folding disorders in heat-stressed rats with histopathological evidence of cardiac injury at 48 hours. Quantitative proteomics analysis by iTRAQ (isobaric tag for relative and absolute quantitation) demonstrated that differential protein expression most closely matched the transcriptomic profile in heat-injured animals at 48 hours. Calculation of protein supersaturation scores supported an increased propensity of proteins to aggregate for proteins that were found to be changing in abundance at 24 hours and in animals with cardiac injury at 48 hours, suggesting a mechanistic association between protein misfolding and the heat-stress response. Conclusions: Pathway analyses at both the transcript and protein levels supported catastrophic deficits in energetics and cellular metabolism and activation of the unfolded protein response in heat-stressed rats with histopathological evidence of persistent heat injury, providing the basis for a systems-level physiological model of heat illness and recovery.

59 BASIC BIOLOGICAL SCIENCES↗

MaxBin: an automated binning method to recover individual genomes from metagenomes using an expectation-maximization algorithm

Recovering individual genomes from metagenomic datasets allows access to uncultivated microbial populations that may have important roles in natural and engineered ecosystems. Understanding the roles of these uncultivated populations has broad application in ecology, evolution, biotechnology and medicine. Accurate binning of assembled metagenomic sequences is an essential step in recovering the genomes and understanding microbial functions. We have developed a binning algorithm, MaxBin, which automates the binning of assembled metagenomic scaffolds using an expectation-maximization algorithm after the assembly of metagenomic sequencing reads. Binning of simulated metagenomic datasets demonstrated that MaxBin had high levels of accuracy in binning microbial genomes. MaxBin was used to recover genomes from metagenomic data obtained through the Human Microbiome Project, which demonstrated its ability to recover genomes from real metagenomic datasets with variable sequencing coverages. Application of MaxBin to metagenomes obtained from microbial consortia adapted to grow on cellulose allowed genomic analysis of new, uncultivated, cellulolytic bacterial populations, including an abundant myxobacterial population distantly related to Sorangium cellulosum that possessed a much smaller genome (5 MB versus 13 to 14 MB) but has a more extensive set of genes for biomass deconstruction. For the cellulolytic consortia, the MaxBin results were compared to binning using emergent self-organizing maps (ESOMs) and differential coverage binning, demonstrating that it performed comparably to these methods but had distinct advantages in automation, resolution of related genomes and sensitivity. The automatic binning software that we developed successfully classifies assembled sequences in metagenomic datasets into recovered individual genomes. The isolation of dozens of species in cellulolytic microbial consortia, including a novel species of myxobacteria that has the smallest genome among all sequenced aerobic myxobacteria, was easily achieved using the binning software. This work demonstrates that the processes required for recovering genomes from assembled metagenomic datasets can be readily automated, an important advance in understanding the metabolic potential of microbes in natural environments. MaxBin is available at https://sourceforge.net/projects/maxbin/.

59 BASIC BIOLOGICAL SCIENCES↗

Community-wide analysis of microbial genome sequence signatures

Background: Analyses of DNA sequences from cultivated microorganisms have revealed genome-wide, taxa-specific nucleotide compositional characteristics, referred to as genome signatures. These signatures have far-reaching implications for understanding genome evolution and potential application in classification of metagenomic sequence fragments. However, little is known regarding the distribution of genome signatures in natural microbial communities or the extent to which environmental factors shape them. Results: We analyzed metagenomic sequence data from two acidophilic biofilm communities, including composite genomes reconstructed for nine archaea, three bacteria, and numerous associated viruses, as well as thousands of unassigned fragments from strain variants and low abundance organisms. Genome signatures, in the form of tetranucleotide frequencies analyzed by emergent self-organizing maps, segregated sequences from all known populations sharing < 50 to 60% average amino acid identity and revealed previously unknown genomic clusters corresponding to low-abundance organisms and a putative plasmid. Signatures were pervasive genome-wide. Clusters were resolved because intra-genome differences resulting from translational selection or protein adaptation to the intracellular (pH ~5) versus extracellular (pH ~1) environment were small relative to inter-genome differences. We found that these genome signatures stem from multiple influences but are primarily manifested through codon composition, which we propose is the result of genome-specific mutational biases. Conclusions: An important conclusion is that shared environmental pressures and interactions among coevolving organisms do not obscure genome signatures in acid mine drainage communities. Thus, genome signatures can be used to assign sequence fragments to populations, an essential prerequisite if metagenomics is to provide ecological and biochemical insights into the functioning of microbial communities.

59 BASIC BIOLOGICAL SCIENCES↗

Sensitivity evaluation of dynamic speckle activity measurements using clustering methods

We evaluate and compare the use of competitive neural networks, self-organizing maps, the expectation-maximization algorithm, K-means, and fuzzy C-means techniques as partitional clustering methods, when the sensitivity of the activity measurement of dynamic speckle images needs to be improved. The temporal history of the acquired intensity generated by each pixel is analyzed in a wavelet decomposition framework, and it is shown that the mean energy of its corresponding wavelet coefficients provides a suited feature space for clustering purposes. The sensitivity obtained by using the evaluated clustering techniques is also compared with the well-known methods of Konishi-Fujii, weighted generalized differences, and wavelet entropy. The performance of the partitional clustering approach is evaluated using simulated dynamic speckle patterns and also experimental data.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A toolkit for detecting technical surprise.

The detection of a scientific or technological surprise within a secretive country or institute is very difficult. The ability to detect such surprises would allow analysts to identify the capabilities that could be a military or economic threat to national security. Sandia's current approach utilizing ThreatView has been successful in revealing potential technological surprises. However, as data sets become larger, it becomes critical to use algorithms as filters along with the visualization environments. Our two-year LDRD had two primary goals. First, we developed a tool, a Self-Organizing Map (SOM), to extend ThreatView and improve our understanding of the issues involved in working with textual data sets. Second, we developed a toolkit for detecting indicators of technical surprise in textual data sets. Our toolkit has been successfully used to perform technology assessments for the Science & Technology Intelligence (S&TI) program.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

Final Technical Report for DE-SC0005467

The objective of this project is to gain a comprehensive understanding of the key atmospheric mechanisms and physical processes associated with temperature extremes in order to better interpret and constrain uncertainty in climate model simulations of future extreme temperatures. To achieve this objective, we first used climate observations and a reanalysis product to identify the key atmospheric circulation patterns associated with extreme temperature days over North America during the late twentieth century. We found that temperature extremes were associated with distinctive signatures in near-surface and mid-tropospheric circulation. The orientations and spatial scales of these circulation anomalies vary with latitude, season, and proximity to important geographic features such as mountains and coastlines. We next examined the associations between daily and monthly temperature extremes and large-scale, recurrent modes of climate variability, including the Pacific-North American (PNA) pattern, the northern annular mode (NAM), and the El Niño-Southern Oscillation (ENSO). The strength of the associations are strongest with the PNA and NAM and weaker for ENSO, and also depend upon season, time scale, and location. The associations are stronger in winter than summer, stronger for monthly than daily extremes, and stronger in the vicinity of the centers of action of the PNA and NAM patterns. In the final stage of this project, we compared climate model simulations of the circulation patterns associated with extreme temperature days over North America with those obtained from observations. Using a variety of metrics and self-organizing maps, we found the multi-model ensemble and the majority of individual models from phase 5 of the Coupled Model Intercomparison Project (CMIP5) generally capture the observed patterns well, including their strength and as well as variations with latitude and season. The results from this project indicate that current models are capable of simulating the large-scale meteorological patterns associated with daily temperature extremes and they suggest that such models can be used to evaluate the extent to which changes in atmospheric circulation will influence future changes in temperature extremes.

54 ENVIRONMENTAL SCIENCES↗

Characterization of the Dynamics of Climate Systems and Identification of Missing Mechanisms Impacting the Long Term Predictive Capabilities of Global Climate Models Utilizing Dynamical Systems Approaches to the Analysis of Observed and Modeled Climate

The goal of this research was to apply fractional and non-linear analysis techniques in order to develop a more complete characterization of climate change and variability for the oceanic, sea ice and atmospheric components of the Earth System. This research applied two measures of dynamical characteristics of time series, the R/S method of calculating the Hurst exponent and Renyi entropy, to observational and modeled climate data in order to evaluate how well climate models capture the long-term dynamics evident in observations. Fractional diffusion analysis was applied to ARGO ocean buoy data to quantify ocean transport. Self organized maps were applied to North Pacific sea level pressure and analyzed in ways to improve seasonal predictability for Alaska fire weather. This body of research shows that these methods can be used to evaluate climate models and shed light on climate mechanisms (i.e., understanding why something happens). With further research, these methods show promise for improving seasonal to longer time scale forecasts of climate.

54 ENVIRONMENTAL SCIENCES↗

Development of Gamma Background Radiation Digital Twin with Machine Learning Algorithms: Application of Unsupervised Machine Learning to Detection of Anomalies and Nuisances in Gamma Background Radiation Environmental Screening Data

Environmental screening of gamma radiation consists of detecting weak nuisance and anomaly signal in the presence of strong and highly varying background. In a typical scenario, a mobile detector-spectrometer continuously measures gamma radiation spectra in short, e.g., one-second, signal acquisition intervals. The measurement data is a 2D matrix, where one dimension is gamma ray energy, and the other dimension is the number of measurements or total time. In principle, gamma radiation sources can be detected and identified from the measured data by their unique spectral lines. Detecting sources from data measured in a search scenario is difficult due to the highly varying background because of naturally occurring radioactive material (NORM), and low signal-to-noise ratio (S/N) of spectral signal measured during one-second acquisition intervals. The objective of this work is to explore unsupervised machine learning (ML) algorithms for development of a digital twin of gamma radiation background, and for detection and identification of weak nuisances and anomalies events in the presence of highly fluctuating background. In one segment of work, we developed a gamma background estimation model using a Longshort term memory (LSTM) network for one-step CPS time series prediction. The LSTM model was validated with two data sets of measurements from two independent NaI detectors positioned on a mobile platform. The data sets contained background radiation only and no orphan isotope sources. The LSTM model was constructed and tested using data from one of the detectors. Performance of the LSTM model was validate through one-step prediction of CPS time series of another NaI detector without re-training. This approach allows to create a digital twin for nuclear background estimation. Using LSTM, it could be possible to detect a source through subtraction of the estimated counts from the measured background. In another segment of work, we investigated detection of gamma emitting sources in the presence of complex background using unsupervised machine learning. Spectral lines of isotopes are difficult to observe in one-second measurements. Averaging over the entire measurement campaign data set reveals spectral lines of most common background isotopes. Spectral lines of orphan sources, which might appear only in a few measurements during the campaign, will be washed out if averaging is performed over the entire measurement data set. The approach we have explored consists of extracting one-second measurements containing weak spectral features through data clustering. Averaging one-second spectra in a cluster should reveal the presence of anomaly sources. We created two ML models using K-means clustering and Neural Network Self-organizing Map (SOM). Performance of these ML models was benchmarked using search data. One data set contained 137 Cs source, and another dataset contained 131 I source.

54 ENVIRONMENTAL SCIENCES↗

Data Mining for Faster, Interpretable Solutions to Inverse Problems:A Case Study Using Additive Manufacturing

Solving inverse problems, where we nd the input values that result in desired values of outputs, can be challenging. The solution process is often computationally expensive and it can be di cult to interpret the solution in high-dimensional input spaces. In this paper, we use a problem from additive manufacturing to address these two issues with the intent of making it easier to solve inverse problems and exploit their results. First, focusing on Gaussian process surrogates that are used to solve inverse problems, we describe how a simple modi cation to the idea of tapering can substantially speed up the surrogate without losing accuracy in prediction. Second, we demonstrate that Kohonen self-organizing maps can be used to visualize and interpret the solution to the inverse problem in the high-dimensional input space. For our data set, as not all input dimensions are equally important, we show that using weighted distances results in a better organized map that makes the relationships among the inputs obvious

97 MATHEMATICS AND COMPUTING↗