Search NASA⌕ Search

SEARCH · Search NASA

Results for “data classification”

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

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

At least 1,585 records · Page 88

Semi-supervised Learning of Dynamical Systems with Neural Ordinary Differential Equations: A Teacher-Student Model Approach

Modeling dynamical systems is crucial for a wide range of tasks, but it remains challenging due to complex nonlinear dynamics, limited observations, or lack of prior knowledge. Recently, data-driven approaches such as Neural Ordinary Differential Equations (NODE) have shown promising results by leveraging the expressive power of neural networks to model unknown dynamics. However, these approaches often suffer from limited labeled training data, leading to poor generalization and suboptimal predictions. On the other hand, semi-supervised algorithms can utilize abundant unlabeled data and have demonstrated good performance in classification and regression tasks. We propose TS-NODE, the first semi-supervised approach to modeling dynamical systems with NODE. TS-NODE explores cheaply generated synthetic pseudo rollouts to broaden exploration in the state space and to tackle the challenges brought by lack of ground-truth system data under a teacher-student model. TS-NODE employs an unified optimization framework that corrects the teacher model based on the student's feedback while mitigating the potential false system dynamics present in pseudo rollouts. TS-NODE demonstrates significant performance improvements over a baseline Neural ODE model on multiple dynamical system modeling tasks.

Wang, Yu↗

STPMiner: A Highperformance Spatiotemporal Pattern Mining Toolbox

The volume of spatiotemporal data being generated from scientific simulations and observations from sensors is growing at an astronomical rate. This data explosion is going to pose three challenges to the existing data mining infrastructure: algorithmic, computational, and I/O. Over the years we have implemented several spatiotemporal data mining algorithms including: outliers/anomalies, colocation patterns, change patterns, clustering, classification, and prediction algorithms. In this paper we briefly discuss the core spatiotemporal pattern mining algorithms along with some of the computational and I/O challenges associated with the big data.

99 GENERAL AND MISCELLANEOUS↗

Landsat D Thematic Mapper image dimensionality reduction and geometric correction accuracy

To characterize and quantify the performance of the Landsat thematic mapper (TM), techniques for dimensionality reduction by linear transformation have been studied and evaluated and the accuracy of the correction of geometric errors in TM images analyzed. Theoretical evaluations and comparisons for existing methods for the design of linear transformation for dimensionality reduction are presented. These methods include the discrete Karhunen Loeve (KL) expansion, Multiple Discriminant Analysis (MDA), Thematic Mapper (TM)-Tasseled Cap Linear Transformation and Singular Value Decomposition (SVD). A unified approach to these design problems is presented in which each method involves optimizing an objective function with respect to the linear transformation matrix. From these studies, four modified methods are proposed. They are referred to as the Space Variant Linear Transformation, the KL Transform-MDA hybrid method, and the First and Second Version of the Weighted MDA method. The modifications involve the assignment of weights to classes to achieve improvements in the class conditional probability of error for classes with high weights. Experimental evaluations of the existing and proposed methods have been performed using the six reflective bands of the TM data. It is shown that in terms of probability of classification error and the percentage of the cumulative eigenvalues, the six reflective bands of the TM data require only a three dimensional feature space. It is shown experimentally as well that for the proposed methods, the classes with high weights have improvements in class conditional probability of error estimates as expected.

Ford, G. E.↗

Searching for pulsars using image pattern recognition

In the modern era of big data, many fields of astronomy are generating huge volumes of data, the analysis of which can sometimes be the limiting factor in research. Fortunately, computer scientists have developed powerful data-mining techniques that can be applied to various fields. In this paper, we present a novel artificial intelligence (AI) program that identifies pulsars from recent surveys by using image pattern recognition with deep neural nets—the PICS (Pulsar Image-based Classification System) AI. The AI mimics human experts and distinguishes pulsars from noise and interference by looking for patterns from candidate plots. Different from other pulsar selection programs that search for expected patterns, the PICS AI is taught the salient features of different pulsars from a set of human-labeled candidates through machine learning. The training candidates are collected from the Pulsar Arecibo L-band Feed Array (PALFA) survey. The information from each pulsar candidate is synthesized in four diagnostic plots, which consist of image data with up to thousands of pixels. The AI takes these data from each candidate as its input and uses thousands of such candidates to train its ∼9000 neurons. The deep neural networks in this AI system grant it superior ability to recognize various types of pulsars as well as their harmonic signals. The trained AI's performance has been validated with a large set of candidates from a different pulsar survey, the Green Bank North Celestial Cap survey. In this completely independent test, the PICS ranked 264 out of 277 pulsar-related candidates, including all 56 previously known pulsars and 208 of their harmonics, in the top 961 (1%) of 90,008 test candidates, missing only 13 harmonics. The first non-pulsar candidate appears at rank 187, following 45 pulsars and 141 harmonics. In other words, 100% of the pulsars were ranked in the top 1% of all candidates, while 80% were ranked higher than any noise or interference. The performance of this system can be improved over time as more training data are accumulated. This AI system has been integrated into the PALFA survey pipeline and has discovered six new pulsars to date.

79 ASTRONOMY AND ASTROPHYSICS↗

Research Data Alliance: Understanding Big Data Analytics Applications in Earth Science

The Research Data Alliance (RDA) enables data to be shared across barriers through focused working groups and interest groups, formed of experts from around the world - from academia, industry and government. Its Big Data Analytics (BDA) interest groups seeks to develop community based recommendations on feasible data analytics approaches to address scientific community needs of utilizing large quantities of data. BDA seeks to analyze different scientific domain applications (e.g. earth science use cases) and their potential use of various big data analytics techniques. These techniques reach from hardware deployment models up to various different algorithms (e.g. machine learning algorithms such as support vector machines for classification). A systematic classification of feasible combinations of analysis algorithms, analytical tools, data and resource characteristics and scientific queries will be covered in these recommendations. This contribution will outline initial parts of such a classification and recommendations in the specific context of the field of Earth Sciences. Given lessons learned and experiences are based on a survey of use cases and also providing insights in a few use cases in detail.

Riedel, Morris↗

A Preliminary Study on the Feasibility of Large Language Models for Detecting Micro-Behaviors Among Team Members in Space Missions

Large-language models (LLMs) have been recently used for spoken language understanding (SLU) to infer meaning and semantics from speech in tasks such as speaker intent and sentiment classification. Due to being trained on large amounts of data, and their ability to understand context and relationships between words, LLMs are competent, enabling them to generalize across tasks without requiring many task-specific training samples. This research examines the feasibility of few-shot learning in LLMs for detecting subtle, brief, and possibly unconscious interactions between team members, called ``micro-behaviors," and provides insights into the appropriate design of LLMs for this task. Our data came from 5 teams participating in a 45-day mission at the US National Aeronautics and Space Administration’s (NASA) Human Exploration Research Analog (HERA). More specifically we used data collected from team interaction battery (TIB) tasks teams performed five times in-mission which comprise an average 1.5 hours of conversation data per day. Micro-behaviors were coded according to an adapted version of Smith & Griffins (2022) theoretical framework in terms of Violation (i.e., presence of valenced behavior, uplifting/positive or discouraging/negative), Intensity (i.e., force of behavior in terms of how uplifting or discouraging is the behavior), and Intent (i.e., motive of the behavior in terms of whether it was deliberate or unintentional). We explore the ability of LLMs to detect the presence and intensity of micro-behaviors. We examine employing and fine-tuning readily available LLMs (i.e., RoBERTa, DistilBERT), as well as prompting state-of-the-art sequence classification models (i.e., Llama-2, Llama-3). In a total of 13,058 conversational turns (17.8% uplifting, 3.3% discouraging, 75.76% neutral, 3.14% nulls), we compute the macro F1-score of the 3-way micro-behavior classification task (i.e., classifying among uplifting, discouraging, and neutral; 33% chance). Results indicate that the RoBERTa model achieves a F1-score of 36.2% (uplift: 43.3% precision (P), 15.1% recall (R); discourage: 20% P, 0.5% R). These results significantly improve when we augment the data via paraphrasing in the RoBERTa model, reaching a 41.2% macro F1-score (uplift: 37.7% P, 86.3% R; discourage: 3.5% P, 1.8% R). Finally, the Llama-2 model with 3-shot prompting yields 38% macro F1-score (uplift: 28.7% P, 20% R; discourage: 7.2% P, 18% R), which is slightly better compared to the RoBERTa model without data augmentation, highlighting the effectiveness of sequence classification models in detecting minority classes with a small sample size. Findings indicate that LLMs hold potential to detect subtle behaviors in conversations, which could be valuable in assessing team behavior in space exploration missions. Future studies will evaluate the performance of different LLM prompting strategies or fine-tuning methods.

Ankush Raut↗

Brown dwarf census with the Dark Energy Survey year 3 data and the thin disc scale height of early L types

In this paper we present a catalogue of 11745 brown dwarfs with spectral types ranging from L0 to T9, photometrically classified using data from the Dark Energy Survey (DES) year 3 release matched to the Vista Hemisphere Survey (VHS) DR3 and Wide-field Infrared Survey Explorer (WISE) data, covering ≈2400 deg 2 up to i AB = 22. The classification method follows the same phototype method previously applied to SDSS-UKIDSS-WISE data. We note the most significant difference comes from the use of DES data instead of SDSS, which allow us to classify almost an order of magnitude more brown dwarfs than any previous search and reaching distances beyond 400 pc for the earliest types. Next, we also present and validate the GalmodBD simulation, which produces brown dwarf number counts as a function of structural parameters with realistic photometric properties of a given survey. We use this simulation to estimate the completeness and purity of our photometric LT catalogue down to i AB = 22, as well as to compare to the observed number of LT types. We put constraints on the thin disc scale height for the early L (L0–L3) population to be around 450 pc, in agreement with previous findings. For completeness, we also publish in a separate table a catalogue of 20863 M dwarfs that passed our colour cut with spectral types greater than M6. Both the LT and the late M catalogues are found at DES release page https://des.ncsa.illinois.edu/releases/other/y3-mlt.

79 ASTRONOMY AND ASTROPHYSICS↗

Adapting ODC for Empirical Evaluation of Pre-Launch Anomalies

This slide presentation reviews the concept of using Orthogonal Defect Classification (ODC) to identify pre-launch anomalies in software. The goals of this work are: (1) To characterize pre-launch software anomalies, using data from multiple spacecraft projects, by means of a defect-analysis technology, Orthogonal Defect Classification (ODC). (2) To support transfer of ODC to NASA projects through applications and demonstrations. Approach: Analyzed anomaly data using adaptation of Orthogonal Defect Classification (ODC) method. This project has adapted ODC for NASA use and applied to NASA projects.

defect analysis↗

Genome sequence of the Thermotoga thermarum type strain (LA 3 T ) from an African solfataric spring

Thermotoga thermarum Windberger et al. 1989 is a member to the genomically well characterized genus Thermotoga in the phylum ' Thermotogae '. T. thermarum is of interest for its origin from a continental solfataric spring vs. predominantly marine oil reservoirs of other members of the genus. The genome of strain LA3T also provides fresh data for the phylogenomic positioning of the (hyper-)thermophilic bacteria. T. thermarum strain LA3 T is the fourth sequenced genome of a type strain from the genus Thermotoga , and the sixth in the family Thermotogaceae to be formally described in a publication. Phylogenetic analyses do not reveal significant discrepancies between the current classification of the group, 16S rRNA gene data and whole-genome sequences. Nevertheless, T. thermarum significantly differs from other Thermotoga species regarding its iron-sulfur cluster synthesis, as it contains only a minimal set of the necessary proteins. Here we describe the features of this organism, together with the complete genome sequence and annotation. The 2,039,943 bp long chromosome with its 2,015 protein-coding and 51 RNA genes is a part of the Genomic Encyclopedia of Bacteria and Archaea project.

59 BASIC BIOLOGICAL SCIENCES↗

Rhode Island Ecological Conservation: Methods for Monitoring Rhode Island Habitats: Contributing to a Framework for Targeted Conservation and Management

Global avian population decline since the 1970s is largely attributable to habitat loss and degradation from anthropogenic disturbances. NASA DEVELOP’s Rhode Island Ecological Conservation team partnered with the Audubon Society of Rhode Island to compute land use land cover (LULC) maps of Rhode Island to aid in the conservation of the state’s 140 bird species. This project aimed to support the partner’s land acquisition strategies with updated and specific LULC classifications showing potential bird-habitat locations across the state. We incorporated remotely sensed data from Landsat 8 and 9 Operational Land Imager (OLI) into LULC maps using unsupervised classification techniques in ArcGIS Pro and supervised classification in Google Earth Engine. We generated six land classifications for 2023, which showed land cover dominated by upland habitats (forests, scrub/shrub, and grasslands), followed by development. We used TerrSet’s Land Change Modeler to forecast LULC change through 2043, using 2011 and 2021 National Land Cover Database (NLCD) land cover maps derived from Landsat 8 and 9 imagery. Project results suggest that non-urban upland and wetland habitats will decrease over time, while development will continue to encroach on non-urban avian habitats. Our maps and associated data will allow for more efficient land acquisition and management efforts to support avian habitat conservation across Rhode Island. Our study shows that data acquisition and processing from open data sources is feasible and further analysis can be done through GIS classification tools. More analysis is needed beyond this study to obtain more detailed land cover maps, though Audubon can aid its targeted conservation efforts with our current, historic, and forecasted LULC maps.

Remote sensing↗

Analysis of Landsat for monitoring vegetables in New York mucklands

This pilot study assessed the feasibility of relying on Landsat multispectral scanner data for inventorying vegetables grown in mucklands,in variably shaped, variably sized fields. Classification of muckland vegetables using a Euclidean distance classifier and a parallelepiped classifier was performed with reasonable accuracy (generally over 60 percent) based on only one date of Landsat data. Prior canonical and principal component analyses did not improve the classification accuracy but did reduce the dimensionality of the data.

Zhu, M. H.↗

Effect of Image Classification Accuracy on Dasymetric Population Estimation

Dasymetric mapping involves the disaggregation of count data, usually relating to population/demographics, from census enumeration areas to smaller target zones with the aid of an ancillary layer related to population density. The ancillary layer is often a binary classification such as developed versus undeveloped, building versus non-building, and residential versus non-residential, in which one class is treated as populated and the other as unpopulated. While dasymetric mapping relies heavily on ancillary data, little research has been done to address the error in ancillary data and its effects on dasymetric mapping accuracy. This chapter reports our research effort to investigate the effect of image classification accuracy on dasymetric population estimates by developing a binomial classification of buildings from high-resolution remote sensor imagery. The classifier was systematically and iteratively manipulated to generate a series of outputs with variegated accuracy characteristics. Lastly, we generated a corresponding series of population estimates based on the mapped building area from each iteration to investigate the relationship between the accuracy of classification and population estimation.

McKee, Jacob↗

Introducing robustness to maximum-likelihood refinement of electron-microsopy data

An expectation-maximization algorithm for maximum-likelihood refinement of electron-microscopy data is presented that is based on finite mixtures of multivariate t-distributions. Compared with the conventionally employed Gaussian mixture model, the t-distribution provides robustness against outliers in the data. An expectation-maximization algorithm for maximum-likelihood refinement of electron-microscopy images is presented that is based on fitting mixtures of multivariate t-distributions. The novel algorithm has intrinsic characteristics for providing robustness against atypical observations in the data, which is illustrated using an experimental test set with artificially generated outliers. Tests on experimental data revealed only minor differences in two-dimensional classifications, while three-dimensional classification with the new algorithm gave stronger elongation factor G density in the corresponding class of a structurally heterogeneous ribosome data set than the conventional algorithm for Gaussian mixtures.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Harmonized Automatic Relay Mitigation of Nefarious Intentional Events (HARMONIE) - Special Protection Scheme (SPS)

The harmonized automatic relay mitigation of nefarious intentional events (HARMONIE) special protection scheme (SPS) was developed to provide adaptive, cyber-physical response to unpredictable disturbances in the electric grid. The HARMONIE-SPS methodology includes a machine learning classification framework that analyzes real time cyber-physical data and determines if the system is in normal conditions, cyber disturbance, physical disturbance, or cyber-physical disturbance. This classification then informs response, if needed and/or suitable, and included cyber-physical corrective actions. Beyond standard power system mitigations, a few novel approaches were developed that included a consensus algorithm-based relay voting scheme, an automated power system triggering condition and corrective action pairing algorithm, and a cyber traffic routing optimization algorithm. Both the classification and response techniques were tested within a newly integrated emulation environment composed of a real-time digital simulator (RTDS) and SCEPTRE™. This report details the HARMONIE-SPS methodology, highlighting both the classification and response techniques, and the subsequent testing results from the emulation environment.

24 POWER TRANSMISSION AND DISTRIBUTION↗

The origin of double-peaked narrow lines in active galactic nuclei. II. Kinematic classifications for the population at z < 0.1

We present optical long-slit observations of the complete sample of 71 Type 2 active galactic nuclei (AGNs) with double-peaked narrow emission lines at z < 0.1 in the Sloan Digital Sky Survey. Double-peaked emission lines are produced by a variety of mechanisms including disk rotation, kiloparsec-scale dual AGNs, and narrow-line region (NLR) kinematics (outflows or inflows). We develop a novel kinematic classification technique to determine the nature of these objects using long-slit spectroscopy alone. We determine that 86% of the double-peaked profiles are produced by moderate-luminosity AGN outflows, 6% are produced by rotation, and 8% are ambiguous. While we are unable to directly identify dual AGNs with long-slit data alone, we explore their potential kinematic classifications with this method. We also find a positive correlation between the NLR size and luminosity of the AGN NLRs (R {sub NLR}∝ L{sub [O III]}{sup 0.21±0.05}), indicating a clumpy two-zone ionization model for the NLR.

79 ASTRONOMY AND ASTROPHYSICS↗

Discrimination of Medicine Radix Astragali from Different Geographic Origins Using Multiple Spectroscopies Combined with Data Fusion Methods

Raman spectra and ultraviolet–visible absorption spectra of four different geographic origins of Radix Astragali were collected. These data were analyzed using kernel principal component analysis combined with sparse representation classification. The results showed that the recognition rate reached 70.44% using Raman spectra for data input and 90.34% using ultraviolet–visible absorption spectra for data input. A new fusion method based on Raman combined with ultraviolet–visible data was investigated and the recognition rate was increased to 96.43%. The experimental results suggested that the proposed data fusion method effectively improved the utilization rate of the original data.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

SU-D-207A-01: Female Pelvic Synthetic CT Generation Based On Joint Shape and Intensity Analysis

Purpose: To develop a method for generating female pelvic synthetic CT (MRCT) images from a single MR scan and evaluate its utility in radiotherapy. Methods: Under IRB-approval, an imaging sequence (T1-VIBE-Dixon) was acquired for 10 patients. This sequence yields 3 useful image volumes of different contrast (“in-phase” T1-weighted, fat and water). A previously published pelvic bone shape model was used to generate a rough bone mask for each patient. A modified fuzzy c-means classification was performed on the multi spectral MR data, with a regularization term that utilizes the prior knowledge provided by the bone mask and addresses the intensity overlap between different tissue types. A weighted sum of classification probabilities with attenuation values yielded MRCT volumes. The mean absolute error (MAE) between MRCT and real CT on various regions was calculated following deformable alignment (Velocity). Intensity modulated Treatment plans based on actual CT and MRCT were made and compared. Results: The average/standard deviation of MAE across 10 patients was 10.1/6.7 HU for muscle, 6.7/4.6 HU for fat, 136.9/53.5 HU for bony tissues under 850 HU (97% of total bone volume), 188.9/119.3 HU for bony tissues above 850 HU and 17.3/13.3 HU for intrapelvic soft tissues. Calculated doses were comparable for plans generated on CT and calculated using MRCT densities or vice versa, with differences in PTV D99% (mean/σ) of (–0.1/0.2 Gy) and (0.3/0.2 Gy), PTV D0.5cc of (–0.3/0.2 Gy) and (–0.4/1.7 Gy). OAR differences were similarly small for comparable structures, with differences in bowel V50Gy of (–0.3/0.2%) and (0.0/0.2%), femur V30Gy of (0.7/1.2%) and (0.2/1.2%), sacrum V20GY of (0.0/0.1%) and (–0.1/1.1%) and mean pelvic V20Gy of (0.0/0.1%) and (0.6/1.8%). Conclusion: MRCT based on a single imaging sequence in the female pelvis is feasible, with acceptably small variations in attenuation estimates and calculated doses to target and critical organs. Work supported by NIHR01EB016079.

60 APPLIED LIFE SCIENCES↗

Application of convolutional neural networks for stellar spectral classification

ABSTRACT Due to the ever-expanding volume of observed spectroscopic data from surveys such as SDSS and LAMOST, it has become important to apply artificial intelligence (AI) techniques for analysing stellar spectra to solve spectral classification and regression problems like the determination of stellar atmospheric parameters Teff, $\rm {\log g}$, and [Fe/H]. We propose an automated approach for the classification of stellar spectra in the optical region using convolutional neural networks (CNNs). Traditional machine learning (ML) methods with ‘shallow’ architecture (usually up to two hidden layers) have been trained for these purposes in the past. However, deep learning methods with a larger number of hidden layers allow the use of finer details in the spectrum which results in improved accuracy and better generalization. Studying finer spectral signatures also enables us to determine accurate differential stellar parameters and find rare objects. We examine various machine and deep learning algorithms like artificial neural networks, Random Forest, and CNN to classify stellar spectra using the Jacoby Atlas, ELODIE, and MILES spectral libraries as training samples. We test the performance of the trained networks on the Indo-U.S. Library of Coudé Feed Stellar Spectra (CFLIB). We show that using CNNs, we are able to lower the error up to 1.23 spectral subclasses as compared to that of two subclasses achieved in the past studies with ML approach. We further apply the trained model to classify stellar spectra retrieved from the SDSS data base with SNR > 20.

Sharma, Kaushal↗