Search NASA⌕ Search

SEARCH · Search NASA

Results for “algorithm selection”

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 1,135 records · Page 63
All sourcesNASA NTRSDOE OSTI
NASA NTRS2003

Software for Partly Automated Recognition of Targets

The Feature Analyst is a computer program for assisted (partially automated) recognition of targets in images. This program was developed to accelerate the processing of high-resolution satellite image data for incorporation into geographic information systems (GIS). This program creates an advanced user interface that embeds proprietary machine-learning algorithms in commercial image-processing and GIS software. A human analyst provides samples of target features from multiple sets of data, then the software develops a data-fusion model that automatically extracts the remaining features from selected sets of data. The program thus leverages the natural ability of humans to recognize objects in complex scenes, without requiring the user to explain the human visual recognition process by means of lengthy software. Two major subprograms are the reactive agent and the thinking agent. The reactive agent strives to quickly learn the user s tendencies while the user is selecting targets and to increase the user s productivity by immediately suggesting the next set of pixels that the user may wish to select. The thinking agent utilizes all available resources, taking as much time as needed, to produce the most accurate autonomous feature-extraction model possible.

Opitz, David↗
NASA NTRS2011

Enhancing a Simple MODIS Cloud Mask Algorithm for the Landsat Data Continuity Mission

The presence of clouds in images acquired by the Landsat series of satellites is usually an undesirable, but generally unavoidable fact. With the emphasis of the program being on land imaging, the suspended liquid/ice particles of which clouds are made of fully or partially obscure the desired observational target. Knowing the amount and location of clouds in a Landsat scene is therefore valuable information for scene selection, for making clear-sky composites from multiple scenes, and for scheduling future acquisitions. The two instruments in the upcoming Landsat Data Continuity Mission (LDCM) will include new channels that will enhance our ability to detect high clouds which are often also thin in the sense that a large fraction of solar radiation can pass through them. This work studies the potential impact of these new channels on enhancing LDCM's cloud detection capabilities compared to previous Landsat missions. We revisit a previously published scheme for cloud detection and add new tests to capture more of the thin clouds that are harder to detect with the more limited arsenal channels. Since there are no Landsat data yet that include the new LDCM channels, we resort to data from another instrument, MODIS, which has these bands, as well as the other bands of LDCM, to test the capabilities of our new algorithm. By comparing our revised scheme's performance against the performance of the official MODIS cloud detection scheme, we conclude that the new scheme performs better than the earlier scheme which was not very good at thin cloud detection.

Wilson, Michael J.↗
NASA NTRS2009

Active Learning with Irrelevant Examples

An improved active learning method has been devised for training data classifiers. One example of a data classifier is the algorithm used by the United States Postal Service since the 1960s to recognize scans of handwritten digits for processing zip codes. Active learning algorithms enable rapid training with minimal investment of time on the part of human experts to provide training examples consisting of correctly classified (labeled) input data. They function by identifying which examples would be most profitable for a human expert to label. The goal is to maximize classifier accuracy while minimizing the number of examples the expert must label. Although there are several well-established methods for active learning, they may not operate well when irrelevant examples are present in the data set. That is, they may select an item for labeling that the expert simply cannot assign to any of the valid classes. In the context of classifying handwritten digits, the irrelevant items may include stray marks, smudges, and mis-scans. Querying the expert about these items results in wasted time or erroneous labels, if the expert is forced to assign the item to one of the valid classes. In contrast, the new algorithm provides a specific mechanism for avoiding querying the irrelevant items. This algorithm has two components: an active learner (which could be a conventional active learning algorithm) and a relevance classifier. The combination of these components yields a method, denoted Relevance Bias, that enables the active learner to avoid querying irrelevant data so as to increase its learning rate and efficiency when irrelevant items are present. The algorithm collects irrelevant data in a set of rejected examples, then trains the relevance classifier to distinguish between labeled (relevant) training examples and the rejected ones. The active learner combines its ranking of the items with the probability that they are relevant to yield a final decision about which item to present to the expert for labeling. Experiments on several data sets have demonstrated that the Relevance Bias approach significantly decreases the number of irrelevant items queried and also accelerates learning speed.

Wagstaff, Kiri↗
NASA NTRS2012

Monitoring Location and Angular Orientation of a Pill

A mobile pill transmitter system moves through, or adjacent to, one or more organs in an animal or human body, while transmitting signals from its present location and/or present angular orientation. The system also provides signals from which the present roll angle of the pill, about a selected axis, can be determined. When the location coordinates angular orientation and the roll angle of the pill are within selected ranges, an aperture on the pill container releases a selected chemical into, or onto, the body. Optionally, the pill, as it moves, provides a sequence of visually perceptible images. The times for image formation may correspond to times at which the pill transmitter system location or image satisfies one of at least four criteria. This invention provides and supplies an algorithm for exact determination of location coordinates and angular orientation coordinates for a mobile pill transmitter (PT), or other similar device that is introduced into, and moves within, a GI tract of a human or animal body. A set of as many as eight nonlinear equations has been developed and applied, relating propagation of a wireless signal between either two, three, or more transmitting antennas located on the PT, to four or more non-coplanar receiving antennas located on a signal receiver appliance worn by the user. The equations are solved exactly, without approximations or iterations, and are applied in several environments: (1) association of a visual image, transmitted by the PT at each of a second sequence of times, with a PT location and PT angular orientation at that time; (2) determination of a position within the body at which a drug or chemical substance or other treatment is to be delivered to a selected portion of the body; (3) monitoring, after delivery, of the effect(s) of administration of the treatment; and (4) determination of one or more positions within the body where provision and examination of a finer-scale image is warranted.

Schipper, John F.↗
NASA NTRS1976

A very-long-baseline interferometer system for geodetic applications

A very-long-baseline interferometer system was designed and built for geodetic applications. Each interferometer terminal records a 360-kHz spectral band of noise from a compact extragalactic radio source. The center frequency of the spectral band can be selected to sample sequentially bands covering a much wider frequency range to obtain subnanosecond accuracy in group-delay measurements. A tunnel-diode pulse generator is used to calibrate the delays in the receiver. The necessary sets of algorithms and computer programs have been developed to analyze the data and have allowed the system to be employed to make accurate determinations of vector baselines, radio-source positions, polar motion, and universal time.

Whitney, A. R.↗
NASA NTRS1981

A global atlas of GEOS-3 significant waveheight data and comparison of the data with national buoy data

The accuracy of the GEOS-3 significant waveheight estimates compared with buoy measurements of significant waveheight were determined. A global atlas of the GEOS-3 significant waveheight estimates gathered is presented. The GEOS-3 significant waveheight estimation algorithm is derived by analyzing the return waveform characteristics of the altimeter. Convergence considerations are examined, the rationale for a smoothing technique is presented and the convergence characteristics of the smoothed estimate are discussed. The GEOS-3 data are selected for comparison with buoy measurements. The GEOS-3 significant waveheight estimates are assembled in the form of a global atlas of contour maps. Both high and low sea state contour maps are presented, and the data are displayed both by seasons and for the entire duration of the GEOS-3 mission.

Mcmillan, J. D.↗
NASA NTRS1983

Real gas flow fields about three dimensional configurations

Real gas, inviscid supersonic flow fields over a three-dimensional configuration are determined using a factored implicit algorithm. Air in chemical equilibrium is considered and its local thermodynamic properties are computed by an equilibrium composition method. Numerical solutions are presented for both real and ideal gases at three different Mach numbers and at two different altitudes. Selected results are illustrated by contour plots and are also tabulated for future reference. Results obtained compare well with existing tabulated numerical solutions and hence validate the solution technique.

Balakrishnan, A.↗
NASA NTRS1991

Scientific requirements for space science data systems

In the 1990's space plasma physics studies will increasingly involve correlative analysis of observations from multiple instruments and multiple spacecraft. The solar terrestrial physics missions in the 1990's will be designed around simultaneous observations from spacecraft monitoring the solar wind, the polar magnetosphere and the near and distant magnetotail. Within these regions, clusters of spacecraft flying in formation will increasingly involve comparative magnetospheric studies. No single lab will have the expertise to process and analyze all of the different types of data so the data repositories will be distributed. Catalog and browse systems will be required to help select events for study. Data compression techniques may be useful in designing the data bases used for selecting events for study. Data compression onboard the spacecraft will be necessary, since instrument data rates will be much larger than available telemetry rates. However, considerable care will be necessary to avoid losing valuable data when applying data compression algorithms.

Walker, Raymond J.↗
NASA NTRS1992

Real time thermal imaging for analysis and control of crystal growth by the Czochralski technique

A real time thermal imaging system with temperature resolution better than +/- 0.5 C and spatial resolution of better than 0.5 mm has been developed. It has been applied to the analysis of melt surface thermal field distributions in both Czochralski and liquid encapsulated Czochralski growth configurations. The sensor can provide single/multiple point thermal information; a multi-pixel averaging algorithm has been developed which permits localized, low noise sensing and display of optical intensity variations at any location in the hot zone as a function of time. Temperature distributions are measured by extraction of data along a user selectable linear pixel array and are simultaneously displayed, as a graphic overlay, on the thermal image.

Wargo, M. J.↗
NASA NTRS1995

Retrieval of water vapor mixing ratios from a laser-based sensor

Langley Research Center has developed a novel external path sensor which monitors water vapor along an optical path between an airplane window and reflective material on the plane's engine. An infrared tunable diode laser is wavelength modulated across a water vapor absorption line at a frequency f. The 2f and DC signals are measured by a detector mounted adjacent to the laser. The 2f/DC ratio depends on the amount of wavelength modulation, the water vapor absorption line being observed, and the temperature, pressure, and water vapor content of the atmosphere. The present work concerns efforts to quantify the contributions of these factors and to derive a method for extracting the water vapor mixing ratio from the measurements. A 3 m cell was fabricated in order to perform laboratory tests of the sensor. Measurements of 2f/DC were made for a series of pressures and modulation amplitudes. During my 1994 faculty fellowship, a computer program was created which allowed 2f/DC to be calculated for any combination of the variables which effect it. This code was used to generate 2f/DC values for the conditions measured in the laboratory. The experimental and theoretical values agreed to within a few percent. As a result, the laser modulation amplitude can now be set in the field by comparing the response of the instrument to the calculated response as a function of modulation amplitude. Once the validity of the computer code was established, it was used to investigate possible candidate absorption lines. 2f/DC values were calculated for pressures, temperatures, and water vapor mixing ratios expected to be encountered in future missions. The results have been incorporated into a database which will be used to select the best line for a particular mission. The database will also be used to select a retrieval technique. For examples under some circumstances there is little temperature dependence in 2f/DC so temperature can be neglected. In other cases, there is a dependence with temperature for a particular pressure, requiring a more complicated retrieval algorithm. Future experimental work is necessary to test agreement with the theoretical values over a range of temperatures and mixing ratios. Additionally, retrieval algorithms for forthcoming missions must be incorporated into the software package which controls the instrument.

Tucker, George F.↗
NASA NTRS1992

Using Fuzzy Logic for Performance Evaluation in Reinforcement Learning

Current reinforcement learning algorithms require long training periods which generally limit their applicability to small size problems. A new architecture is described which uses fuzzy rules to initialize its two neural networks: a neural network for performance evaluation and another for action selection. This architecture is applied to control of dynamic systems and it is demonstrated that it is possible to start with an approximate prior knowledge and learn to refine it through experiments using reinforcement learning.

Berenji, Hamid R.↗
NASA NTRS1998

The Mars Surveyor 2001 Lander: A First Step Toward Precision Landing

As the robotic exploration of Mars continues, science objectives have driven mission and flight system development towards the use of precision landing technology such that small surface features, such as craters, can be investigated. In addition, the surface rendezvous elements of human exploration missions will require landing accuracy that is greatly improved over that achievable with ballistic flight. With improved approach navigation and hypersonic maneuvering technologies, the MSP'01 Lander is taking the first significant step toward precision landing on Mars. This advance requires both the ability to generate lift during the atmospheric flight and an on-board guidance algorithm to direct a three-axis control system. Many configuration options were examined to generate the required lift, with an afterbody-mounted deployable flap emerging as the lightest-weight solution. Five candidate guidance algorithms have been developed and submitted to the MSP'01 Project for evaluation. Through high-fidelity simulation, each of these algorithms has demonstrated the ability to greatly improve upon the landed accuracy provided by ballistic flight. As a result, the science community should expect to be within 10 km of the specified landing target. In fact, depending on the selected aeroshell L/D, a 5-km precision landing goal is achievable with greater than 90% confidence.

Braun, Robert D.↗
NASA NTRSDate not supplied

Guidance Enhancements and Performance Assessments for the Mars Ascent Vehicle Spin-Stabilized Upper Stage Configuration

he objective of the Mars Sample Return (MSR) campaign is to return samples from the surface of Mars to Earth for research. As one element of the MSR campaign, the Mars Ascent Vehicle (MAV) is responsible for transporting the samples from the surface of Mars to a Low-Martian Orbit (LMO) for retrieval. Complete autonomy is required throughout ascent, and orbital insertion is constrained by tight dispersion boundaries. An unguided, spin-stabilized second stage for MAV has been selected over a guided upper-stage to drive mass savings and reduce overall MSR campaign mass risk, at the cost of reduced GNC capability. To address this design change, the MAV GNC team has derived a robust prediction algorithm, building on previous energy management schemes, that solves for a single inertial pointing direction solution for the spin-stabilized 2nd stage burn. Algorithm stability is explored that compared to previous versions of the algorithm. Also, a set of analytical partials was developed to study MAV’s dispersed orbital insertion performance with respect to MAV system uncertainties. These partials were verified through simulation analysis and prove useful for analytical insight into the dynamics of MAV during the 2nd stage maneuver.

Jason M Everett↗
NASA NTRSDate not supplied

Guidance Enhancements and Performance Assessments for the Mars Ascent Vehicle Spin-Stabilized Upper Stage Configuration

The objective of the Mars Sample Return (MSR) campaign is to return samples from the surface of Mars to Earth for research. As one element of the MSR campaign, the Mars Ascent Vehicle (MAV) is responsible for transporting the samples from the surface of Mars to a Low-Martian Orbit (LMO) for retrieval. Complete autonomy is required throughout ascent, and orbital insertion is constrained by tight dispersion boundaries. An unguided, spin-stabilized second stage for MAV has been selected over a guided upper-stage to drive mass savings and reduce overall MSR campaign mass risk, at the cost of reduced GNC capability. To address this design change, the MAV GNC team has derived a robust prediction algorithm, building on previous energy management schemes, that solves for a single inertial pointing direction solution for the spin-stabilized 2nd stage burn. Algorithm stability is explored that compared to previous versions of the algorithm. Also, a set of analytical partials was developed to study MAV’s dispersed orbital insertion performance with respect to MAV system uncertainties. These partials were verified through simulation analysis and prove useful for analytical insight into the dynamics of MAV during the 2nd stage maneuver.

GNC↗
NASA NTRS2006

Active Learning with Irrelevant Examples

Active learning algorithms attempt to accelerate the learning process by requesting labels for the most informative items first. In real-world problems, however, there may exist unlabeled items that are irrelevant to the user's classification goals. Queries about these points slow down learning because they provide no information about the problem of interest. We have observed that when irrelevant items are present, active learning can perform worse than random selection, requiring more time (queries) to achieve the same level of accuracy. Therefore, we propose a novel approach, Relevance Bias, in which the active learner combines its default selection heuristic with the output of a simultaneously trained relevance classifier to favor items that are likely to be both informative and relevant. In our experiments on a real-world problem and two benchmark datasets, the Relevance Bias approach significantly improved the learning rate of three different active learning approaches.

machine learning↗
NASA NTRS2014

Elastic Model Transitions: A Hybrid Approach Utilizing Quadratic Inequality Constrained Least Squares (LSQI) and Direct Shape Mapping (DSM)

A method for transitioning linear time invariant (LTI) models in time varying simulation is proposed that utilizes a hybrid approach for determining physical displacements by augmenting the original quadratically constrained least squares (LSQI) algorithm with Direct Shape Mapping (DSM) and modifying the energy constraints. The approach presented is applicable to simulation of the elastic behavior of launch vehicles and other structures that utilize discrete LTI finite element model (FEM) derived mode sets (eigenvalues and eigenvectors) that are propagated throughout time. The time invariant nature of the elastic data presents a problem of how to properly transition elastic states from the prior to the new model while preserving motion across the transition and ensuring there is no truncation or excitation of the system. A previous approach utilizes a LSQI algorithm with an energy constraint to effect smooth transitions between eigenvector sets with no requirement that the models be of similar dimension or have any correlation. This approach assumes energy is conserved across the transition, which results in significant non-physical transients due to changing quasi-steady state energy between mode sets, a phenomenon seen when utilizing a truncated mode set. The computational burden of simulating a full mode set is significant so a subset of modes is often selected to reduce run time. As a result of this truncation, energy between mode sets may not be constant and solutions across transitions could produce non-physical transients. In an effort to abate these transients an improved methodology was developed based on the aforementioned approach, but this new approach can handle significant changes in energy across mode set transitions. It is proposed that physical velocities due to elastic behavior be solved for using the LSQI algorithm, but solve for displacements using a two-step process that independently addresses the quasi-steady-state and non-steady-state contributions to the elastic displacement. For structures subject to large external forces, such as thrust or atmospheric drag, it is imperative to capture these forces when solving for elastic displacement. To simplify the mathematical formulation, assumptions are made regarding mass matrix normalization, constant external forcing, and constant viscous damping. These simplifications allow for direct solutions to the quasi-steady-state displacements through a process titled Direct Shape Mapping. DSM solves for the displacements using the eigenvalues of the elastic modes and the external forcing and returns a set of elastic displacements dictated by the eigenvectors of the post-transition mode set. For the non-steady-state contributions to displacement we formulate a LSQI problem that is constrained by energy of the non-steady state terms. The contributions from the quasi-steady-state and non-steady state solutions are then combined to obtain the physical displacements associated with the new set of eigenvectors. Results for the LSQI-DSM approach show significant reduction/complete removal of transients across mode set transitions while maintaining elastic motion from the prior state. For time propagation applications employing discrete elastic models that need to be transitioned in time and where running with full a full mode set is not feasible, the method developed offers a practical solution to simulating vehicle elasticity.

Hannan, Mike R.↗
NASA NTRSDate not supplied

Adaptive Learning for Reliability Analysis using Support Vector Machines

A novel algorithm is presented for adaptive learning of an unknown function that separates two regions of a domain.In the context of reliability analysis these two regions represent the failure domain, where a set of constraints or requirements are violated, and a safe domain where they are satisfied. The Limit State Function (LSF) separates these two regions. Evaluating the constraints for a given parameter point requires the evaluation of a computational model that may well be expensive. For this reason we wish to construct a meta-model that can estimate the LSFas accurately as possible, using only a limited amount of training data. This work presents an adaptive strategy employing a Support Vector Machine (SVM) as a meta-model to provide a semi-algebraic approximation of the LSF.We describe an optimization process that is used to select informative parameter points to add to training data at each iteration to improve the accuracy of this approximation. A formulation is introduced for bounding the predictions of the meta-model; in this way we seek to incorporate this aspect of Gaussian Process Models (GPMs) within anSVM meta-model. Finally, we apply our algorithm to two benchmark test cases, demonstrating performance that is comparable with, if not superior, to a standard technique for reliability analysis that employs GPMs

Adaptive learning↗
NASA NTRS2015

Resolving Ice Cloud Optical Thickness Biases Between CALIOP and MODIS Using Infrared Retrievals

Despite its importance as one of the key radiative properties that determines the impact of upper tropospheric clouds on the radiation balance, ice cloud optical thickness (IOT) has proven to be one of the more challenging properties to retrieve from space-based remote sensing measurements. In particular, optically thin upper tropospheric ice clouds (cirrus) have been especially challenging due to their tenuous nature, extensive spatial scales, and complex particle shapes and light scattering characteristics. The lack of independent validation motivates the investigation presented in this paper, wherein systematic biases between MODIS Collection 5 (C5) and CALIOP Version 3 (V3) unconstrained retrievals of tenuous IOT (< 3) are examined using a month of collocated A-Train observations. An initial comparison revealed a factor of two bias between the MODIS and CALIOP IOT retrievals. This bias is investigated using an infrared (IR) radiative closure approach that compares both products with MODIS IR cirrus retrievals developed for this assessment. The analysis finds that both the MODIS C5 and the unconstrained CALIOP V3 retrievals are biased (high and low, respectively) relative to the IR IOT retrievals. Based on this finding, the MODIS and CALIOP algorithms are investigated with the goal of explaining and minimizing the biases relative to the IR. For MODIS we find that the assumed ice single scattering properties used for the C5 retrievals are not consistent with the mean IR COT distribution. The C5 ice scattering database results in the asymmetry parameter (g) varying as a function of effective radius with mean values that are too large. The MODIS retrievals have been brought into agreement with the IR by adopting a new ice scattering model for Collection 6 (C6) consisting of a modified gamma distribution comprised of a single habit (severely roughened aggregated columns); the C6 ice cloud optical property models have a constant g approx. = 0.75 in the mid-visible spectrum, 5-15% smaller than C5. For CALIOP, the assumed lidar ratio for unconstrained retrievals is fixed at 25 sr for the V3 data products.This value is found to be inconsistent with the constrained (predominantly nighttime) CALIOP retrievals. An experimental data set was produced using a modified lidar ratio of 32 sr for the unconstrained retrievals (an increase of 28%), selected to provide consistency with the constrained V3 results. These modifications greatly improve the agreement with the IR and provide consistency between the MODIS and CALIOP products. Based on these results the recently released MODIS C6 optical products use the single habit distribution given above, while the upcoming CALIOP V4 unconstrained algorithm will use higher lidar ratios for unconstrained retrievals.

Cloud↗