Search NASA⌕ Search

SEARCH · Search NASA

Results for “false data”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,063 records · Page 59

KEPLER-20: A SUN-LIKE STAR WITH THREE SUB-NEPTUNE EXOPLANETS AND TWO EARTH-SIZE CANDIDATES

We present the discovery of the Kepler-20 planetary system, which we initially identified through the detection of five distinct periodic transit signals in the Kepler light curve of the host star 2MASS J19104752+4220194. From high-resolution spectroscopy of the star, we find a stellar effective temperature T{sub eff} = 5455 {+-} 100 K, a metallicity of [Fe/H] = 0.01 {+-} 0.04, and a surface gravity of log g = 4.4 {+-} 0.1. We combine these estimates with an estimate of the stellar density derived from the transit light curves to deduce a stellar mass of M{sub *} = 0.912 {+-} 0.034 M{sub Sun} and a stellar radius of R{sub *} = 0.944{sup +0.060}{sub -0.095} R{sub Sun }. For three of the transit signals, we demonstrate that our results strongly disfavor the possibility that these result from astrophysical false positives. We accomplish this by first identifying the subset of stellar blends that reproduce the precise shape of the light curve and then using the constraints on the presence of additional stars from high angular resolution imaging, photometric colors, and the absence of a secondary component in our spectroscopic observations. We conclude that the planetary scenario is more likely than that of an astrophysical false positive by a factor of 2 Multiplication-Sign 10{sup 5} (Kepler-20b), 1 Multiplication-Sign 10{sup 5} (Kepler-20c), and 1.1 Multiplication-Sign 10{sup 3} (Kepler-20d), sufficient to validate these objects as planetary companions. For Kepler-20c and Kepler-20d, the blend scenario is independently disfavored by the achromaticity of the transit: from Spitzer data gathered at 4.5 {mu}m, we infer a ratio of the planetary to stellar radii of 0.075 {+-} 0.015 (Kepler-20c) and 0.065 {+-} 0.011 (Kepler-20d), consistent with each of the depths measured in the Kepler optical bandpass. We determine the orbital periods and physical radii of the three confirmed planets to be 3.70 days and 1.91{sup +0.12}{sub -0.21} R{sub Circled-Plus} for Kepler-20b, 10.85 days and 3.07{sup +0.20}{sub -0.31} R{sub Circled-Plus} for Kepler-20c, and 77.61 days and 2.75{sup +0.17}{sub -0.30} R{sub Circled-Plus} for Kepler-20d. From multi-epoch radial velocities, we determine the masses of Kepler-20b and Kepler-20c to be 8.7 {+-} 2.2 M{sub Circled-Plus} and 16.1 {+-} 3.5 M{sub Circled-Plus }, respectively, and we place an upper limit on the mass of Kepler-20d of 20.1 M{sub Circled-Plus} (2{sigma}).

79 ASTRONOMY AND ASTROPHYSICS↗

Imaging Modes for Ground Penetrating Radar and Their Relation to Detection Performance

The focus of this paper is an empirical study conducted to determine how imaging modes for ground penetrating radar (GPR) affect buried object detection performance. GPR data was collected repeatedly over lanes whose buried objects were mostly non-metallic. This data was collected and processed with a GPR antenna array, system hardware, and processing software developed by the authors and their colleagues. The system enables GPR data to be collected, imaged, and processed in real-time on a moving vehicle. The images are focused by applying multistatic and synthetic aperture imaging techniques either separately or jointly to signal scans acquired by the GPR antenna array. An image-based detection statistic derived from the ratio of buried object energy in the foreground to energy of soil in the background is proposed. Detection – false alarm performance improved significantly when the detection algorithm was applied to focused multistatic synthetic aperture radar (SAR) images rather than to unfocused GPR signal scans.

47 OTHER INSTRUMENTATION↗

Random forest models accurately classify synthetic opioids using high-dimensionality mass spectrometry datasets

Detection of novel threat agents presents several challenges, a principle one being the development of untargeted methods to screen an increasing number of threat chemicals whose exact structures are unknown. With the use of Machine Learning (ML) tools, we can guide the development of analytical methods for broad-spectrum detection of unbounded threat chemical families in complex mixtures. Toward this goal, we used nominal mass and high-resolution mass spectrometry data for hundreds of synthetic opioids and non-opioid compounds. We tested two ML techniques, logistic regression and random forest, to develop models towards a practical, implementable method for opioid detection. We found that of these tested ML methods, random forest models resulted in the highest validation accuracy (95+%) for both nominal mass and high-resolution classification of opioids versus non-opioids, with low false positive and false negative rates. The RF models were then used to successfully predict the classification of 10 compounds—five opioids and five non-opioids not part of the training and validation analysis. This application of ML is a critical step towards the development of field-deployable nominal mass spectrometers with ML-driven analyses for classification of emergent threats.

Arasteh, Kourosh [Lawrence Livermore National Labo↗

A data-driven sensor placement approach for detecting voltage violations in distribution systems

Stochastic fluctuations in power injections from distributed energy resources (DERs) combined with load variability can cause constraint violations (e.g., exceeded voltage limits) in electric distribution systems. To monitor grid operations, sensors are placed to measure important quantities such as the voltage magnitudes. Here, in this paper, we consider a sensor placement problem which seeks to identify locations for installing sensors that can capture all possible violations of voltage magnitude limits. We formulate a bilevel optimization problem that minimizes the number of sensors and avoids false sensor alarms in the upper level while ensuring detection of any voltage violations in the lower level. This problem is challenging due to the nonlinearity of the power flow equations and the presence of binary variables. Accordingly, we employ recently developed conservative linear approximations of the power flow equations that overestimate or underestimate the voltage magnitudes. By replacing the nonlinear power flow equations with conservative linear approximations, we can ensure that the resulting sensor locations and thresholds are sufficient to identify any constraint violations. Additionally, we apply various problem reformulations to significantly improve computational tractability while simultaneously ensuring an appropriate placement of sensors. Lastly, we improve the quality of the results via an approximate gradient descent method that adjusts the sensor thresholds. We demonstrate the effectiveness of our proposed method for several test cases, including a system with multiple switching configurations.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Biochemical Detection and Identification False Alarm Rate Dependence on Wavelength Using Laser Induced Fluorescence

Most organic and many inorganic materials absorb strongly in specific wavelength ranges in the deep UV between about 220nm and 300nm. Excitation within these absorption bands results in native fluorescence emission. Each compound or composite material, such as a bacterial spore, has a unique excitation-emission fingerprint that can be used to provide information about the material. The sensitivity and specificity with which these materials can be detected and identified depends on the excitation wavelength and the number and location of observation wavelengths.We will present data on our deep ultraviolet Targeted Ultraviolet Chemical Sensors that demonstrate the sensitivity and specificity of the sensors. In particular, we will demonstrate the ability to quantitatively differentiate a wide range of biochemical agent targets against a wide range of background materials. We will describe the relationship between spectral resolution and specificity in target identification, as well as simple, fast, algorithms to identify materials.Hand-held, battery operated instruments using a deep UV laser and multi-band detection have been developed and deployed on missions to the Antarctic, the Arctic, and the deep ocean with the capability of detecting a single bacterial spore and to differentiate a wide range of organic and biological compounds.

native flourescence↗

Global Freeze/Thaw Product from L-Band Radiometer Data

The NASA Soil Moisture Active Passive (SMAP) mission has been successfully operated for almost three years. The SMAP freeze/thaw algorithm is based on a seasonal threshold approach. It’s important to have a stable and self-consistent freeze and thaw reference that can be applied for multi-year dataset. The three-year long radiometer datasets allow us to reassess the criteria of reference setup and evaluate its stability. In this paper, we first refined the freeze reference requirements and compare three different methods of setting up the thaw reference to minimize the false flags. The original freeze/thaw products is in the polar grid and only cover the region north of 45° N latitude. The limitation is due to lack of enough freezing days in the lower latitude, where the freezing reference cannot be generated. To extend the freeze/thaw product to global region, we combine the single channel algorithm in the lower latitude and southern atmosphere. The global results have been validated through WMO air temperature.

Colliander, Andreas↗

TRANSITING PLANETS WITH LSST. II. PERIOD DETECTION OF PLANETS ORBITING 1 M{sub ⊙} HOSTS

The Large Synoptic Survey Telescope (LSST) will photometrically monitor ∼10{sup 9} stars for 10 years. The resulting light curves can be used to detect transiting exoplanets. In particular, as demonstrated by Lund et al., LSST will probe stellar populations currently undersampled in most exoplanet transit surveys, including out to extragalactic distances. In this paper we test the efficiency of the box-fitting least-squares (BLS) algorithm for accurately recovering the periods of transiting exoplanets using simulated LSST data. We model planets with a range of radii orbiting a solar-mass star at a distance of 7 kpc, with orbital periods ranging from 0.5 to 20 days. We find that standard-cadence LSST observations will be able to reliably recover the periods of Hot Jupiters with periods shorter than ∼3 days; however, it will remain a challenge to confidently distinguish these transiting planets from false positives. At the same time, we find that the LSST deep-drilling cadence is extremely powerful: the BLS algorithm successfully recovers at least 30% of sub-Saturn-size exoplanets with orbital periods as long as 20 days, and a simple BLS power criterion robustly distinguishes ∼98% of these from photometric (i.e., statistical) false positives.

79 ASTRONOMY AND ASTROPHYSICS↗

Inflation after false vacuum decay: new evidence from BICEP2

Last year we argued that if slow-roll inflation followed the decay of a false vacuum in a large landscape, the steepening of the scalar potential between the inflationary plateau and the barrier generically leads to a potentially observable suppression of the scalar power spectrum at large distances. Here we revisit this analysis in light of the recent BICEP2 results. Assuming that both the BICEP2 B-mode signal and the Planck analysis of temperature fluctuations hold up, we find that the data now discriminate more sharply between our scenario and ΛCDM. Nonzero tensor modes exclude standard ΛCDM with notable but not yet conclusive confidence: at ∼3.8σ if r = 0.2, or at ∼ 3.5σ if r = 0.15. Of the two steepening models of our previous work, one is now ruled out by existing bounds on spatial curvature. The other entirely reconciles the tension between BICEP2 and Planck. Upcoming EE polarization measurements have the potential to rule out unmodified ΛCDM decisively. Next generation Large Scale Structure surveys can further increase the significance. More precise measurements of BB at low ℓ will help distinguish our scenario from other explanations. If steepening is confirmed, the prospects for detecting open curvature increase but need not be large.

79 ASTRONOMY AND ASTROPHYSICS↗

Quantifying the Performance of Charge-Coupled Devices in Ambient Conditions - Oral presentation [Slides]

Telescope surveys have given us a great deal of information about our universe, but the images they capture carry with them an inherent limitation. The question then is how do we take this information to the next level? The answer: the Dark Energy Spectroscopic Instrument (DESI). DESI is an instrument that will measure the distance to tens of millions of galaxies in our night sky. This information can be combined with already existing images to construct a three dimensional map of our universe providing a great deal of new opportunities for cosmological research. The DESI guidance system consists of 10 detectors called charge-coupled devices (CCDs). Each CCD is made of silicon atoms that emit electrons when struck with light, the electrons are counted and then used to reconstruct an image. But, CCDs suffer from an issue known as ‘dark current’ which are false counts that come from thermal motions of the silicon atoms. This is particularly problematic since they contribute to the uncertainty of a measurement without contributing to our signal. This causes a drop in the signal to noise ratio, a value that needs to be maximized in order to meet DESI’s high precision requirements. This summer was spent ensuring the DESI guidance system would meet its specifications. Data was collected using a CCD of the same type that would be used on DESI and the effectiveness of dark current removal was tested. Exposures were taken for a wide range of temperatures and exposure lengths and a number of dark current removal methods were implemented. While further testing is required, the initial results are quite promising and the DESI guidance system is on track to meet its specifications

47 OTHER INSTRUMENTATION↗

Toward Intelligent Software Defect Detection

Source code level software defect detection has gone from state of the art to a software engineering best practice. Automated code analysis tools streamline many of the aspects of formal code inspections but have the drawback of being difficult to construct and either prone to false positives or severely limited in the set of defects that can be detected. Machine learning technology provides the promise of learning software defects by example, easing construction of detectors and broadening the range of defects that can be found. Pinpointing software defects with the same level of granularity as prominent source code analysis tools distinguishes this research from past efforts, which focused on analyzing software engineering metrics data with granularity limited to that of a particular function rather than a line of code.

Benson, Markland J.↗

Neutron Skin Measurement of 208Pb and 48Ca Using Parity Violating Electron Scattering

Parity-violating electron scattering experiments (PVES) provides a clean probe of neutron densities that is model independent and free from strong interaction uncertainties in interpretation. The PREX-2 and CREX experiments were run in 2019 and 2020 at Jefferson laboratory measured the nucleon skin thickness, the difference between the r.m.s. neutron radius Rn and the r.m.s. proton radius Rp, of 208Pb and 48Ca via parity violating electroweak asymmetry in the elastic scattering of longitudinally polarized electrons. PREX-2 experiment was performed with 950 MeV electrons scattered at a 5¿ angle with Q2 = 0.00616 ± 0.00004 (GeV/c)2, while CREX used 2182 MeV electrons at the same angle with Q2 = 0.0297 ± 0.0002 (GeV/c)2. For PREX-2 the measured asymmetry was AP V = 550 ± 16 (stat.) ± (syst.) ppb, which corresponds to Rskin = 0.278 ± 0.078 (exp.) ± 0.012 (theo.) fm. The CREX asymmetry was AP V = 2668 ± 106 (stat.) ± 40 (syst.) ppb, which corresponds to Rskin = 0.121 ± 0.026 (exp.) ±0.024 (model) fm. One of the crucial systematic uncertainty that PREX-2 and CREX were sensitive to was the non-parity violating asymmetries that resulted from the helicity correlated false asymmetries in the polarized electron beam. There was a lot of work put towards understanding and suppressing the asymmetries arising from these effects. The parity violating asymmetry measurement required a very precise determination of the electron beam polarization. To accurately determine the beam polarization, a Compton polarimeter was used during both PREX-2 and CREX. A careful alignment of the laser to the Fabry-Perot cavity, data analysis and systematic control was employed to get a precise beam polarization result for the experiments. The PREX-2 measurement has broad implications for increasing our knowledge about neutron star structure and the equation of state of nuclear matter. The combined PREX and CREX results will have implications for future energy density functional calculations and the theory of nuclear structure.

Premathilake, Sachinthani↗

LU60645GT and MA132843GT Catalogues of Lunar and Martian Impact Craters Developed Using a Crater Shape-based Interpolation Crater Detection Algorithm for Topography Data

For Mars, 57,633 craters from the manually assembled catalogues and 72,668 additional craters identified using several crater detection algorithms (CDAs) have been merged into the MA130301GT catalogue. By contrast, for the Moon the most complete previous catalogue contains only 14,923 craters. Two recent missions provided higher-quality digital elevation maps (DEMs): SELENE (in 1/16° resolution) and Lunar Reconnaissance Orbiter (we used up to 1/512°). This was the main motivation for work on the new Crater Shape-based interpolation module, which improves previous CDA as follows: (1) it decreases the number of false-detections for the required number of true detections; (2) it improves detection capabilities for very small craters; and (3) it provides more accurate automated measurements of craters' properties. The results are: (1) LU60645GT, which is currently the most complete (up to ∼D>=8 km) catalogue of Lunar craters; and (2) MA132843GT catalogue of Martian craters complete up to ∼D>=2 km, which is the extension of the previous MA130301GT catalogue. As previously achieved for Mars, LU60645GT provides all properties that were provided by the previous Lunar catalogues, plus: (1) correlation between morphological descriptors from used catalogues; (2) correlation between manually assigned attributes and automated measurements; (3) average errors and their standard deviations for manually and automatically assigned attributes such as position coordinates, diameter, depth/diameter ratio, etc; and (4) a review of positional accuracy of used datasets. Additionally, surface dating could potentially be improved with the exhaustiveness of this new catalogue. The accompanying results are: (1) the possibility of comparing a large number of Lunar and Martian craters, of e.g. depth/diameter ratio and 2D profiles; (2) utilisation of a method for re-projection of datasets and catalogues, which is very useful for craters that are very close to poles; and (3) the extension of the previous framework for evaluation of CDAs with datasets and ground-truth catalogue for the Moon.

Craters↗

Computing single step operators of logic programming in radial basis function neural networks

Logic programming is the process that leads from an original formulation of a computing problem to executable programs. A normal logic program consists of a finite set of clauses. A valuation I of logic programming is a mapping from ground atoms to false or true. The single step operator of any logic programming is defined as a function (T{sub p}:I→I). Logic programming is well-suited to building the artificial intelligence systems. In this study, we established a new technique to compute the single step operators of logic programming in the radial basis function neural networks. To do that, we proposed a new technique to generate the training data sets of single step operators. The training data sets are used to build the neural networks. We used the recurrent radial basis function neural networks to get to the steady state (the fixed point of the operators). To improve the performance of the neural networks, we used the particle swarm optimization algorithm to train the networks.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Detecting Peer-to-Peer Botnets in SCADA Systems

Supervisory Control and Data Acquisition (SCADA) systems monitor and control critical infrastructure such as the smart grid. As SCADA systems become increasingly interconnected and adopt more and more cyber-enabled components, the risks of cyber attacks become a major concern. Due to their decentralized organization, peer-to-peer (P2P) botnets are resilient to many existing takedown measures and can be exploited as an effective way to launch cyber attacks on SCADA systems. However, little work has been done to detect P2P botnets in SCADA systems, which carry traffic flows with characteristics significantly different from the Internet. In this paper, we design a P2P-botnet detection method for SCADA systems, leveraging built-in traffic monitoring capabilities of SCADA networking devices. The proposed method consists of two stages. In the first stage, we design a simple feature test to filter out non-P2P hosts, which significantly reduces the data volume for P2P-botnet identification. In the second stage, we jointly consider flow-based and connectivity-based features that effectively set apart bots from benign hosts. We propose to use unsupervised learning for P2P-botnet identification, which not only identifies known P2P botnets but also captures newly emerged ones. Our simulation results show that the proposed system achieves high detection rates with very few false positives. Furthermore, our evaluation shows that the proposed method can detect hosts running P2P SCADA applications that are infected by P2P bots.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Computer aided detection of surgical retained foreign object for prevention

Purpose: Surgical retained foreign objects (RFOs) have significant morbidity and mortality. They are associated with approximately $1.5 × 10{sup 9} annually in preventable medical costs. The detection accuracy of radiographs for RFOs is a mediocre 59%. The authors address the RFO problem with two complementary technologies: a three-dimensional (3D) gossypiboma micro tag, the μTag that improves the visibility of RFOs on radiographs, and a computer aided detection (CAD) system that detects the μTag. It is desirable for the CAD system to operate in a high specificity mode in the operating room (OR) and function as a first reader for the surgeon. This allows for fast point of care results and seamless workflow integration. The CAD system can also operate in a high sensitivity mode as a second reader for the radiologist to ensure the highest possible detection accuracy. Methods: The 3D geometry of the μTag produces a similar two dimensional (2D) depiction on radiographs regardless of its orientation in the human body and ensures accurate detection by a radiologist and the CAD. The authors created a data set of 1800 cadaver images with the 3D μTag and other common man-made surgical objects positioned randomly. A total of 1061 cadaver images contained a single μTag and the remaining 739 were without μTag. A radiologist marked the location of the μTag using an in-house developed graphical user interface. The data set was partitioned into three independent subsets: a training set, a validation set, and a test set, consisting of 540, 560, and 700 images, respectively. A CAD system with modules that included preprocessing μTag enhancement, labeling, segmentation, feature analysis, classification, and detection was developed. The CAD system was developed using the training and the validation sets. Results: On the training set, the CAD achieved 81.5% sensitivity with 0.014 false positives (FPs) per image in a high specificity mode for the surgeons in the OR and 96.1% sensitivity with 0.81 FPs per image in a high sensitivity mode for the radiologists. On the independent test set, the CAD achieved 79.5% sensitivity with 0.003 FPs per image in a high specificity mode for the surgeons and 90.2% sensitivity with 0.23 FPs per image in a high sensitivity mode for the radiologists. Conclusions: To the best of the authors’ knowledge, this is the first time a 3D μTag is used to produce a recognizable, substantially similar 2D projection on radiographs regardless of orientation in space. It is the first time a CAD system is used to search for man-made objects over anatomic background. The CAD system for the μTags achieved reasonable performance in both the high specificity and the high sensitivity modes.

60 APPLIED LIFE SCIENCES↗

TH-AB-207A-12: CT Lung Cancer Screening and the Effects of Further Dose Reduction On CAD Performance

Purpose: CT lung screening is already performed at low doses. In this study, we investigated the effects of further dose reduction on a lung-nodule CAD detection algorithm. Methods: The original raw CT data and images from 348 patients were obtained from our local database of National Lung Screening Trial (NLST) cases. 61 patients (17.5%) had at least one nodule reported on the NLST reader forms. All scans were acquired with fixed mAs (25 for standard-sized patients, 40 for large patients) on a 64-slice scanner (Sensation 64, Siemens Healthcare). All images were reconstructed with 1-mm slice thickness, B50 kernel. Based on a previously-published technique, we added noise to the raw data to simulate reduced-dose versions of each case at 50% and 25% of the original NLST dose (i.e. approximately 1.0 and 0.5 mGy CTDIvol). For each case at each dose level, a CAD detection algorithm was run and nodules greater than 4 mm in diameter were reported. These CAD results were compared to “truth”, defined as the approximate nodule centroids from the NLST forms. Sensitivities and false-positive rates (FPR) were calculated for each dose level, with a sub-analysis by nodule LungRADS category. Results: For larger category 4 nodules, median sensitivities were 100% at all three dose levels, and mean sensitivity decreased with dose. For the more challenging category 2 and 3 nodules, the dose dependence was less obvious. Overall, mean subject-level sensitivity varied from 38.5% at 100% dose to 40.4% at 50% dose, a difference of only 1.9%. However, median FPR quadrupled from 1 per case at 100% dose to 4 per case at 25% dose. Conclusions: Dose reduction affected nodule detectability differently depending on the LungRADS category, and FPR was very sensitive at sub-screening levels. Care should be taken to adapt CAD for the very challenging noise characteristics of screening. Funding support: NIH U01 CA181156; Disclosures (McNitt-Gray): Institutional research agreement, Siemens Healthcare; Past recipient, research grant support, Siemens Healthcare; Consultant, Toshiba America Medical Systems; Consultant, Samsung Electronics.

60 APPLIED LIFE SCIENCES↗

All digital precision processing of ERTS images

The author has identified the following significant results. Experimentation was conducted to evaluate the performance of the Sequential Similarity Detection Algorithm (SSDA) to detect and locate ground central points (GCP) automatically using MSS data. Recent experiments with ERTS data having a temporal separation of from 17 to 72 days between the search area and the GCP have shown that the algorithm can find the GCP's and with an overall probability of 88%. Band 5 appears to give the best results. A modified reseau detection algorithm has been applied to 2 RBV scenes separated by a 12 day period. The algorithm correctly located all 486 reseaus. No false reseaus were located in a companion experiment. Changes in apparent reseau position, due to camera characteristics, were never greater than 3 picture elements in either axis. The positional error of a geometrically corrected image has been predicted by the use of an APL program. The maximum deviation of the GCP's from true UTM coordinate position was computed to be 190 meters. The RMS positional error of all GCP's was 106 meters. Further refinement of the algorithm is expected to reduce the errors.

Bernstein, R.↗

Interpretable Machine Learning for Molecular Biosignatures: a Novel Single-Sample Feature Importance Method That Is Sensitive To Statistical Interactions

Isotope ratio mass spectrometry (IRMS) of volatiles (e.g., CO 2 ) promises to be a powerful tool for potential biosignature detection for future missions to ocean worlds (OW) such as Europa and Enceladus. Machine learning (ML) methods for IRMS data could enable science autonomy by onboard prediction of seawater chemistry and biosignature presence. However, ML models are likely to be complex and involve statistical interactions between features (variables), which can make predictions seem opaque and enigmatic. For ML predictions as significant as extraterrestrial biosignatures, we must place extraordinary confidence in models. It is therefore essential that these models make interpretable predictions (i.e., human-understandable) and include false-prediction diagnostics. We achieve high accuracy and interpretability in ML biosignature and seawater chemistry models for OW through a nearest-neighbors feature selection tool that detects statistical interactions between predictors, constructs interaction networks for visualization of selected features working together to make a prediction, and reports single-sample feature importance scores for false-detection diagnostics. Here we develop a novel single-sample nearest-neighbors projected distance regression(ssNPDR) feature selection method that improves upon existing single-sample algorithms through the inclusion of statistical interactions while providing false-prediction diagnostics for ML models.

geochemistry↗