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 1,243 records · Page 69

Controlling basins of attraction in a neural network-based telemetry monitor

The size of the basins of attraction around fixed points in recurrent neural nets (NNs) can be modified by a training process. Controlling these attractive regions by presenting training data with various amount of noise added to the prototype signal vectors is discussed. Application of this technique to signal processing results in a classification system whose sensitivity can be controlled. This new technique is applied to the classification of temporal sequences in telemetry data.

Bell, Benjamin↗

Data Processing And Machine Learning Methods For Multi-Modal Operator State Classification Systems

This document is intended as an introduction to a set of common signal processing learning methods that may be used in the software portion of a functional crew state monitoring system. This includes overviews of both the theory of the methods involved, as well as examples of implementation. Practical considerations are discussed for implementing modular, flexible, and scalable processing and classification software for a multi-modal, multi-channel monitoring system. Example source code is also given for all of the discussed processing and classification methods.

Machine learning↗

Data Science and the Knowledge Discovery Adventure

This talk will cover the important steps involved in the data science and knowledge discovery process: • Initial fact gathering (interview domain experts, review reports, articles, state-of-the-art) • Identify the problem (prediction, classification, statistical analysis, etc.) • Survey supporting data sources • Understand the data (numerical, categorical, text, sampling rate, data quality issues, etc.) • Selecting relevant features and sources • Acquire the data (set up agreements with the data stewards, APIs to download, etc.) • Merge data sources (temporal, spatial, common key, other ontologies...) • Feature Engineering (non linear domain knowledge or physics-based relationships) • Build data processing pipeline (may need to tap into data stream, develop parallel processing algorithm, federated learning etc.) • Build model and test (tune hyper-parameters, cross validation.) • Analyze/Validate results (do the results make sense. Does it answer the original question). • Deploy/Publish (Monitor and assess benefits)

Data science↗

Use of Integrated Instrumentation to Detect and Classify Targets in Shallow Water

The detection and classification of targets in shallow water environments (e.g., depth < 20 m) is of increasing interest, both in security applications and more general environmental studies. Here, we report the ability of a cabled integrated instrumentation system to detect and classify targets in a shallow (15 m deep), narrow (150 m wide) channel influenced by moderate tidal currents (peak currents < 2 m/s). The instrumentation system incorporates a passive acoustic array, multibeam sonar, acoustic camera, stereo-optical camera, and acoustic Doppler current profiler. The system effectiveness is benchmarked using autonomous “cooperative” targets (e.g., drifters with known position and acoustic characteristics), divers, and opportunistic observations of fish and marine mammals. By fusing data streams from multiple sensors, detection and classification rates are improved, particularly for smaller targets. Approaches to data management (data rates exceed 80 MB/s), real-time classification, and de-noising of acoustic imagery are also discussed.

16 TIDAL AND WAVE POWER↗

Automated Detection of Anomalous Shipping Manifests to Identify Illicit Trade

We describe an approach to analyzing trade data which uses clustering to detect similarities across shipping manifest records, classification to evaluate clustering results and categorize new unseen shipping data records, and visual analytics to provide to support situation awareness in dynamic decision making to monitor and warn against the movement of radiological threat materials through search, analysis and forecasting capabilities. The evaluation of clustering results through classification and systematic inspection of the clusters show the clusters have strong semantic cohesion and offer novel ways to detect transactions related to nuclear smuggling.

classification↗

Predictive Big Data Analytics: A Study of Parkinson’s Disease Using Large, Complex, Heterogeneous, Incongruent, Multi-Source and Incomplete Observations

Background A unique archive of Big Data on Parkinson’s Disease is collected, managed and disseminated by the Parkinson’s Progression Markers Initiative (PPMI). The integration of such complex and heterogeneous Big Data from multiple sources offers unparalleled opportunities to study the early stages of prevalent neurodegenerative processes, track their progression and quickly identify the efficacies of alternative treatments. Many previous human and animal studies have examined the relationship of Parkinson’s disease (PD) risk to trauma, genetics, environment, co-morbidities, or life style. The defining characteristics of Big Data–large size, incongruency, incompleteness, complexity, multiplicity of scales, and heterogeneity of information-generating sources–all pose challenges to the classical techniques for data management, processing, visualization and interpretation. We propose, implement, test and validate complementary model-based and model-free approaches for PD classification and prediction. To explore PD risk using Big Data methodology, we jointly processed complex PPMI imaging, genetics, clinical and demographic data. Methods and Findings Collective representation of the multi-source data facilitates the aggregation and harmonization of complex data elements. This enables joint modeling of the complete data, leading to the development of Big Data analytics, predictive synthesis, and statistical validation. Using heterogeneous PPMI data, we developed a comprehensive protocol for end-to-end data characterization, manipulation, processing, cleaning, analysis and validation. Specifically, we (i) introduce methods for rebalancing imbalanced cohorts, (ii) utilize a wide spectrum of classification methods to generate consistent and powerful phenotypic predictions, and (iii) generate reproducible machine-learning based classification that enables the reporting of model parameters and diagnostic forecasting based on new data. We evaluated several complementary model-based predictive approaches, which failed to generate accurate and reliable diagnostic predictions. However, the results of several machine-learning based classification methods indicated significant power to predict Parkinson’s disease in the PPMI subjects (consistent accuracy, sensitivity, and specificity exceeding 96%, confirmed using statistical n-fold cross-validation). Clinical (e.g., Unified Parkinson's Disease Rating Scale (UPDRS) scores), demographic (e.g., age), genetics (e.g., rs34637584, chr12), and derived neuroimaging biomarker (e.g., cerebellum shape index) data all contributed to the predictive analytics and diagnostic forecasting. Conclusions Model-free Big Data machine learning-based classification methods (e.g., adaptive boosting, support vector machines) can outperform model-based techniques in terms of predictive precision and reliability (e.g., forecasting patient diagnosis). We observed that statistical rebalancing of cohort sizes yields better discrimination of group differences, specifically for predictive analytics based on heterogeneous and incomplete PPMI data. UPDRS scores play a critical role in predicting diagnosis, which is expected based on the clinical definition of Parkinson’s disease. Even without longitudinal UPDRS data, however, the accuracy of model-free machine learning based classification is over 80%. The methods, software and protocols developed here are openly shared and can be employed to study other neurodegenerative disorders (e.g., Alzheimer’s, Huntington’s, amyotrophic lateral sclerosis), as well as for other predictive Big Data analytics applications.

Science & Technology - Other Topics↗

Ground water use inventory in Minnesota using LANDSAT data

An irrigation ground water use inventory in Sherburne County, Minnesota, was attempted by indirectly locating irrigation wells by determining the location of irrigated lands. Digital classification of LANDSAT MSS imagery was employed using the IDIMS at Goddard Space Flight Center. Single-date classification proved moderately successful after accurate ground truth data were obtained. A combination of supervised and unsupervised classification was used. Multitemporal analysis using August 10 and May 21, 1978 imagery was also attempted. This unsupervised classification proved most successful in identifying crops and irrigated fields.

Beissel, D. R.↗

Relative Accuracy Assessment of LANDSAT-4 MSS and TM Data for Level 1 Land Cover Inventory

Digital data for the Washington, DC scene simultaneously acquired by the LANDSAT-4 Multispectral Scanner (MSS) and the LANDSAT-4 thematic mapper (TM) was compared. Classification success for the TM and MSS data sets was determined by a per pixel comparison with digitized ground verification data (GVD). These GVD were comprised of Level 7 land cover (developed, agriculture, forest, water, wetlands, and barren) for four USGS 7.5 minute topographic quadrangle maps. The relative improvement in classification success for TM was between 11% and 14%, or about a factor of 1.3, for these data. This represents a meaningful improvement in accuracy for Level 7 land cover categorization for TM relative to MSS, particularly when errors of omission and commission were considered.

Middleton, E. M.↗

A systematic comparison of genome-scale clustering algorithms

Background: A wealth of clustering algorithms has been applied to gene co-expression experiments. These algorithms cover a broad range of approaches, from conventional techniques such as k-means and hierarchical clustering, to graphical approaches such as k-clique communities, weighted gene co-expression networks (WGCNA) and paraclique. Comparison of these methods to evaluate their relative effectiveness provides guidance to algorithm selection, development and implementation. Most prior work on comparative clustering evaluation has focused on parametric methods. Graph theoretical methods are recent additions to the tool set for the global analysis and decomposition of microarray co-expression matrices that have not generally been included in earlier methodological comparisons. In the present study, a variety of parametric and graph theoretical clustering algorithms are compared using well-characterized transcriptomic data at a genome scale from Saccharomyces cerevisiae. Methods: For each clustering method under study, a variety of parameters were tested. Jaccard similarity was used to measure each cluster’s agreement with every GO and KEGG annotation set, and the highest Jaccard score was assigned to the cluster. Clusters were grouped into small, medium, and large bins, and the Jaccard score of the top five scoring clusters in each bin were averaged and reported as the best average top 5 (BAT5) score for the particular method. Results: Clusters produced by each method were evaluated based upon the positive match to known pathways. This produces a readily interpretable ranking of the relative effectiveness of clustering on the genes. Methods were also tested to determine whether they were able to identify clusters consistent with those identified by other clustering methods. Conclusions: Validation of clusters against known gene classifications demonstrate that for this data, graph-based techniques outperform conventional clustering approaches, suggesting that further development and application of combinatorial strategies is warranted.

59 BASIC BIOLOGICAL SCIENCES↗

SEP Time-to-Maximum (TTM) Studies and the Ionic Charge States of Solar Heavy Ions

1. We published a report on the TTM method and presented the results at the 28 International Cosmic Ray Conference. Included in this paper were our results on the energy-dependent charge state of Fe in the 200 1 April 15 ground- level event, the largest of Cycle 23. We showed that our results agreed well with direct measurements from SAMPEX in this event. We also presented results for the 2001 April 14 event, one of the largest impulsive events of Cycle 23. 2. We extracted TTM values for protons and ions of various energies in approx. 50 SEP events in Cycle 23, using time profiles measured by ACE, Wind, GOES, IMPS, SOHO, and SAMPEX. This event sample included many large gradual events, as well as several impulsive events. The TTM analyses typically yielded approx. 300 datapoints per event. 3. We used these data to make preliminary event classifications according to: (a) shapes of the rigidity and/or speed dependence of TTMs in each event; (b) behavior of proton TTMs compared to that of heavier ions; and (c) the behavior of Fe TTMs compared to that of lighter Z greater than I ions. 4. We cross-checked proton fluence and TTM values from 8 instruments on 6 satellites, so as to understand the origin of systematic instrument-to-instrument discrepancies. By also comparing the observed proton spectra, it was possible to extract correction factors to the nominal proton energy bins that removed or greatly reduced the instrument-to-instrument spread among proton TTMs. 5. We began comparisons of proton and He-4 TTM values in some particularly well measured, large SEP events measured by IMP8 in Cycles 21 and 22 when the instruments onboard were functioning better. It is hoped that these data will provide a better understanding of event-to-event variation in the functional dependence of TTMs on velocity and rigidity.

Tylka, Allan J.↗

Flood-plain delineation using multispectral data analysis

The paper explores the application of a remote sensing technique that may permit the determination of flood-plain areas without the extensive work associated with existing techniques. Multispectral scanner data were simulated by utilizing the density differences in a color-infrared transparency for a section of the Navasota River, Texas. The transparency was taken from a low-flying aircraft and covered an area approximating a square mile. The simulated data were processed by an automatic classification technique previously developed in the remote sensing field. The technique used involves the application of the maximum likelihood rule in order to categorize the data being processed. An attempt was made to distinguish between areas known to be in the flood plain and those outside. A reasonabke correlation was found between boundaries based on computer-processed multispectral data and those produced by techniques currently in use.

Harker, G. R.↗

Physical properties of asteroids

The physical properties of asteroids were studied by telescopic observations and laboratory and theoretical work. Spectrophotometry from 0.3 to 1.1 microns and 1.2, 1.6 and 2.2 micron photometry allow spectral-compositional classification of asteroids. Based on laboratory data and telescopic observations, it was found that infrared measurements at 1.2, 1.6 and 2.2 microns provide a relatively rapid and accurate method for the classification of minor planets and are important in comparing asteroids with meteorites. This technique was proven and employed in an expanded survey of Apollo-Amor-Aten and other unusual asteroids recently scanned by IRAS.

Veeder, G. J.↗

Remote sensing of earth terrain

A systematic approach for the identification of terrain media such as vegetation canopy, forest, and snow covered fields is developed using the optimum polarimetric classifier. The covariance matrices for the various terrain cover are computed from theoretical models of random medium by evaluating the full polarimetric scattering matrix elements. The optimal classification scheme makes use of a quadratic distance measure and is applied to classify a vegetation canopy consisting of both trees and grass. Experimentally measured data are used to validate the classification scheme. Theoretical probability of classification error using the full polarimetric matrix are compared with classification based on single features including the phase difference between the VV and HH polarization returns. It is shown that the full polarimetric results are optimal and provide better classification performance than single feature measurements.

Kong, J. A.↗

Remote sensing of earth terrain

A systematic approach for the identification of terrain media such as vegetation canopy, forest, and snow covered fields is developed using the optimum polarimetric classifier. The covariance matrices for the various terrain covers are computed from the theoretical models of random medium by evaluating the full polarimetric scattering matrix elements. The optimal classification scheme makes use of a quadratic distance measure and is applied to classify a vegetation canopy consisting of both trees and grass. Experimentally measured data are used to validate the classification scheme. Theoretical probability of classification error using the full polarimetric matrix are compared with classification based on single features including the phase difference between the VV and HH polarization returns. It is shown that the full polarimetric results are optimal and provide better classification performance than single feature measurements. A systematic approach is presented for obtaining the optimal polarimetric matched filter which produces maximum contrast between two scattering classes, each represented by its respective covariance matrix.

Kong, J. A.↗

Identification of terrain cover using the optimum polarimetric classifier

A systematic approach for the identification of terrain media such as vegetation canopy, forest, and snow-covered fields is developed using the optimum polarimetric classifier. The covariance matrices for various terrain cover are computed from theoretical models of random medium by evaluating the scattering matrix elements. The optimal classification scheme makes use of a quadratic distance measure and is applied to classify a vegetation canopy consisting of both trees and grass. Experimentally measured data are used to validate the classification scheme. Analytical and Monte Carlo simulated classification errors using the fully polarimetric feature vector are compared with classification based on single features which include the phase difference between the VV and HH polarization returns. It is shown that the full polarimetric results are optimal and provide better classification performance than single feature measurements.

Kong, J. A.↗

Recommended data sets, corn segments and spring wheat segments, for use in program development

The sets of Large Area Crop Inventory Experiment sites, crop year 1978, which are recommended for use in the development and evaluation of classification techniques based on LANDSAT spectral data are presented. For each site, the following exists: (1) accuracy assessment digitized ground truth; (2) a minimum of 5 percent of the scene ground truth identified as corn or spring wheat; and (3) at least four acquisitions of acceptable data quality during the growing season of the crop of interest. The recommended data sets consist of 41 corn/soybean sites and 17 spring wheat sites.

Austin, W. W.↗

Pattern-recognition techniques applied to performance monitoring of the DSS 13 34-meter antenna control assembly

The results of applying pattern recognition techniques to diagnose fault conditions in the pointing system of one of the Deep Space network's large antennas, the DSS 13 34-meter structure, are discussed. A previous article described an experiment whereby a neural network technique was used to identify fault classes by using data obtained from a simulation model of the Deep Space Network (DSN) 70-meter antenna system. Described here is the extension of these classification techniques to the analysis of real data from the field. The general architecture and philosophy of an autonomous monitoring paradigm is described and classification results are discussed and analyzed in this context. Key features of this approach include a probabilistic time-varying context model, the effective integration of signal processing and system identification techniques with pattern recognition algorithms, and the ability to calibrate the system given limited amounts of training data. Reported here are recognition accuracies in the 97 to 98 percent range for the particular fault classes included in the experiments.

Mellstrom, J. A.↗