Search NASA⌕ Search

SEARCH · Search NASA

Results for “Neural Networks (Computer)”

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 469 records · Page 26
All sourcesNASA NTRSDOE OSTI
NASA NTRS1992

Spectral analysis for automated exploration and sample acquisition

Future space exploration missions will rely heavily on the use of complex instrument data for determining the geologic, chemical, and elemental character of planetary surfaces. One important instrument is the imaging spectrometer, which collects complete images in multiple discrete wavelengths in the visible and infrared regions of the spectrum. Extensive computational effort is required to extract information from such high-dimensional data. A hierarchical classification scheme allows multispectral data to be analyzed for purposes of mineral classification while limiting the overall computational requirements. The hierarchical classifier exploits the tunability of a new type of imaging spectrometer which is based on an acousto-optic tunable filter. This spectrometer collects a complete image in each wavelength passband without spatial scanning. It may be programmed to scan through a range of wavelengths or to collect only specific bands for data analysis. Spectral classification activities employ artificial neural networks, trained to recognize a number of mineral classes. Analysis of the trained networks has proven useful in determining which subsets of spectral bands should be employed at each step of the hierarchical classifier. The network classifiers are capable of recognizing all mineral types which were included in the training set. In addition, the major components of many mineral mixtures can also be recognized. This capability may prove useful for a system designed to evaluate data in a strange environment where details of the mineral composition are not known in advance.

Eberlein, Susan↗
NASA NTRS1995

Highly-Parallel, Highly-Compact Computing Structures Implemented in Nanotechnology

In this paper, we describe work in which we are evaluating how the evolving properties of nano-electronic devices could best be utilized in highly parallel computing structures. Because of their combination of high performance, low power, and extreme compactness, such structures would have obvious applications in spaceborne environments, both for general mission control and for on-board data analysis. However, the anticipated properties of nano-devices mean that the optimum architecture for such systems is by no means certain. Candidates include single instruction multiple datastream (SIMD) arrays, neural networks, and multiple instruction multiple datastream (MIMD) assemblies.

Crawley, D. G.↗
NASA NTRS2003

Analysis of Moessbauer Data from Mars: A Database and Artificial Neural Network for Identification of Iron-bearing Phases

The exploration of the planet Mars is one of the major goals within the Solar system exploration programs of the US-American space agency NASA and the European Space Agency ESA. In particular the search for water and life and understanding of the history of the surface and atmosphere will be the major tasks of the upcoming space missions to Mars. The miniaturized Moessbauer spectrometer MIMOS II has been selected for the NASA Mars-Exploration-Rover twin-mission to Mars in 2003 and the ESA 2003 Mars-Express Beagle 2 mission. Reduced in size and weight, in comparison to ordinary laboratory setup, the sensor head just weights approximately 400 g, with a volume of (50x50x90) cu mm, and holds two gamma-ray sources: the stronger for experiments and the weaker for calibrations. The collimator (in sample direction) also shields the primary radiation off the detectors. Around the drive four detectors are mounted. The detectors are made of Si-PIN-photodiodes in chip form (100 sq mm, thickness of 0.5 mm). The control unit is located in a separate electronics board. This board is responsible for the power supply, generation of the drive's velocity reference signal, read of the detector pulses to record the spectrum, data storage and communication with the host computer. After more than four decades from the discovery of the Moessbauer effect, more than 400 minerals were studied at different temperatures. Their Moessbauer parameters were reported in the literature, and have been recently collected in a data bank. Previous Mars-missions, namely Viking and Mars Pathfinder, revealed Si, Al, Fe, Mg, Ca, K, Ti, S and Cl to be the major constituents in soil and rock elemental composition of the red planet. More than 200 minerals already studied by Moessbauer spectroscopy contain significant amounts of these elements. A considerable number of Moessbauer studies were also carried out on meteorites and on Moon samples. Looking backward in the studies of the whole Moessbauer community, we have built a specific library containing Moessbauer parameters of those possible Mars minerals. The selected minerals, their Moessbauer parameter values (min. max. s.d and number of available data), main site substitution, behavior as a function of temperature and a ranking as expected to be found on Mars were organized. Mars-analogue Fe-bearing minerals not studied by Moessbauer spectroscopy are being collected and investigated. In addition, it an identification system based on Artificial Neural Networks (ANN) was implemented which enables fast and precise mineral identification from the experimental Moessbauer parameters at a given temperature.

P A de Souza Jr.↗
NASA NTRS1999

Information Technology and the Autonomous Control of a Mars In-Situ Propellant Production System

With the rapidly increasing performance of information technology, i.e., computer hardware and software systems, as well as networks and communication systems, a new capability is being developed that holds the clear promise of greatly increased exploration capability, along with dramatically reduced design, development, and operating costs. These new intelligent systems technologies, utilizing knowledge-based software and very high performance computer systems, will provide new design and development tools, scheduling mechanisms, and vehicle and system health monitoring capabilities. In addition, specific technologies such as neural nets will provide a degree of machine intelligence and associated autonomy which has previously been unavailable to the mission and spacecraft designer and to the system operator. One of the most promising applications of these new information technologies is to the area of in situ resource utilization. Useful resources such as oxygen, compressed carbon dioxide, water, methane, and buffer gases can be extracted and/or generated from planetary atmospheres, such as the Martian atmosphere. These products, when used for propulsion and life-support needs can provide significant savings in the launch mass and costs for both robotic and crewed missions. In the longer term the utilization of indigenous resources is an enabling technology that is vital to sustaining long duration human presence on Mars. This paper will present the concepts that are currently under investigation and development for mining the Martian atmosphere, such as temperature-swing adsorption, zirconia electrolysis etc., to create propellants and life-support materials. This description will be followed by an analysis of the information technology and control needs for the reliable and autonomous operation of such processing plants in a fault tolerant manner, as well as the approach being taken for the development of the controlling software. Finally, there will be a brief discussion of the verification and validation process so crucial to the implementation of mission-critical software.

Gross, Anthony R.↗
NASA NTRS2020

Amino Acid Encoding for Deep Learning Applications

Background: The number of applications of deep learning algorithms in bioinformatics is increasing as they usually achieve superior performance over classical approaches, especially, when bigger training datasets are available. In deep learning applications, discrete data, e.g. words or n-grams in language, or amino acids or nucleotides in bioinformatics, are generally represented as a continuous vector through an embedding matrix. Recently, learning this embedding matrix directly from the data as part of the continuous iteration of the model to optimize the target prediction – a process called ‘end-to-end learning’ – has led to state-ofthe-art results in many fields. Although usage of embeddings is well described in the bioinformatics literature, the potential of end-to-end learning for single amino acids, as compared to more classical manually-curated encoding strategies, has not been systematically addressed. To this end, we compared classical encoding matrices, namely one-hot, VHSE8 and BLOSUM62, to end-to-end learning of amino acid embeddings for two different prediction tasks using three widely used architectures, namely recurrent neural networks (RNN), convolutional neural networks (CNN), and the hybrid CNN-RNN. Results: By using different deep learning architectures, we show that end-to-end learning is on par with classical encodings for embeddings of the same dimension even when limited training data is available, and might allow for a reduction in the embedding dimension without performance loss, which is critical when deploying the models to devices with limited computational capacities. We found that the embedding dimension is a major factor in controlling the model performance. Surprisingly, we observed that deep learning models are capable of learning from random vectors of appropriate dimension. Conclusion: Our study shows that end-to-end learning is a flexible and powerful method for amino acid encoding. Further, due to the flexibility of deep learning systems, amino acid encoding schemes should be benchmarked against random vectors of the same dimension to disentangle the information content provided by the encoding scheme from the distinguishability effect provided by the scheme.

Hesham ElAbd↗
NASA NTRS1997

Applying Space Technology to Enhance Control of an Artificial Arm

At the present time, myoelectric prostheses perform only one function of the hand: open and close with the thumb, index and middle finger coming together to grasp various shaped objects. To better understand the limitations of the current single-function prostheses and the needs of the individuals who use them, The Institute for Rehabilitation and Research (TIRR), sponsored by the National Institutes of Health (August 1992 - November 1994), surveyed approximately 2500 individuals with upper limb loss. When asked to identify specific features of their current electric prosthesis that needed improvement, the survey respondents overwhelmingly identified the lack of wrist and finger movement as well as poor control capability. Simply building a mechanism with individual finger and wrist motion is not enough. Individuals with upper limb loss tend to reject prostheses that require continuous visual monitoring and concentration to control. Robotics researchers at NASA's Johnson Space Center (JSC) and Rice University have made substantial progress in myoelectric teleoperation. A myoelectric teleoperation system translates signals generated by an able-bodied robot operator's muscles during hand motions into commands that drive a robot's hand through identical motions. Farry's early work in myoelectric teleoperation used variations over time in the myoelectric spectrum as inputs to neural networks to discriminate grasp types and thumb motions. The resulting schemes yielded up to 93% correct classification on thumb motions. More recently, Fernandez achieved 100% correct non-realtime classification of thumb abduction, extension, and flexion on the same myoelectric data. Fernandez used genetic programming to develop functions that discriminate between thumb motions using myoelectric signal parameters. Genetic programming (GP) is an evolutionary programming method where the computer can modify the discriminating functions' form to improve its performance, not just adjust numerical coefficients or weights. Although the function development may require much computational time and many training cases, the resulting discrimination functions can run in realtime on modest computers. These results suggest that myoelectric signals might be a feasible teleoperation medium, allowing an operator to use his or her own hand and arm as a master to intuitively control an anthropomorphic robot in a remote location such as outer space.

Atkins, Diane↗
NASA NTRS2018

Learning Model Structural Uncertainty with Gaussian Processes

The advent of commercially available quantum computers has marked the beginning of quantum computing as a reality. Both quantum gate and annealing computers have been released by major computer hardware companies. In this work, the D-Wave 2XTM quantum annealing computer housed at the NASA Advanced Systems computational facility is investigated to accelerate Machine Learning (ML) for image registration. NASA collects large amounts of images over the globe remotely using space-based monitoring. Images of a fixed areas of the land surface are taken over time. Due to the orbit of the sensors, the viewing angles deviate slightly, and it is necessary to align or register the images precisely to create image time series over the land surface. Unaligned images can lead to substantial analysis errors. These time-series are then used in modeling Earth Systems models such as hydrological, weather, and carbon monitoring models. In this work, we consider the Moderate Resolution Image Spectrometer (MODIS) data collected by the NASA's terra satellite. Artificial Neural Networks (ANNs) is a natural fit for ML modelling of images. Several successes have been reported using machine learning related to image processing. We investigate the use of ML to register MODIS images. ANNs are investigated in combination with a Restricted Boltzmann Machines (RBM) as an auto-encoder. We will present results showing the accuracy and efficiency of this approach.The D-Wave 2XTM quantum annealer samples the ground-state wave-function of a spin-Ising systems with quadratic interactions between qubits and a Chimera connectivity. The system sits in a ~15 mK thermal bath. One can think of the system as being placed in the ground state initially and subject to thermal excitations governed by Boltzmann statistics. If this is assumed true, one can use the statistics from the D-Wave 2XTM to train RBMs. Generating statistics for training Boltzmann machines is an NP-hard problem and constitutes the largest compute cost. We investigate the use of the D-Wave 2XTM to accelerate the training of the RBMs in our ANNs and report on the results.

Kouatchou, Jules↗
NASA NTRS1990

Highly parallel computation

Highly parallel computing architectures are the only means to achieve the computation rates demanded by advanced scientific problems. A decade of research has demonstrated the feasibility of such machines and current research focuses on which architectures designated as multiple instruction multiple datastream (MIMD) and single instruction multiple datastream (SIMD) have produced the best results to date; neither shows a decisive advantage for most near-homogeneous scientific problems. For scientific problems with many dissimilar parts, more speculative architectures such as neural networks or data flow may be needed.

Denning, Peter J.↗
NASA NTRS2012

Heading Toward Launch with the Integrated Multi-Satellite Retrievals for GPM (IMERG)

The Day-l algorithm for computing combined precipitation estimates in GPM is the Integrated Multi-satellitE Retrievals for GPM (IMERG). We plan for the period of record to encompass both the TRMM and GPM eras, and the coverage to extend to fully global as experience is gained in the difficult high-latitude environment. IMERG is being developed as a unified U.S. algorithm that takes advantage of strengths in the three groups that are contributing expertise: 1) the TRMM Multi-satellite Precipitation Analysis (TMPA), which addresses inter-satellite calibration of precipitation estimates and monthly scale combination of satellite and gauge analyses; 2) the CPC Morphing algorithm with Kalman Filtering (KF-CMORPH), which provides quality-weighted time interpolation of precipitation patterns following cloud motion; and 3) the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks using a Cloud Classification System (PERSIANN-CCS), which provides a neural-network-based scheme for generating microwave-calibrated precipitation estimates from geosynchronous infrared brightness temperatures. In this talk we summarize the major building blocks and important design issues driven by user needs and practical data issues. One concept being pioneered by the IMERG team is that the code system should produce estimates for the same time period but at different latencies to support the requirements of different groups of users. Another user requirement is that all these runs must be reprocessed as new IMERG versions are introduced. IMERG's status at meeting time will be summarized, and the processing scenario in the transition from TRMM to GPM will be laid out. Initially, IMERG will be run with TRMM-based calibration, and then a conversion to a GPM-based calibration will be employed after the GPM sensor products are validated. A complete reprocessing will be computed, which will complete the transition from TMPA.

Huffman, George J.↗
NASA NTRS1997

Smart-Pixel Array Processors Based on Optimal Cellular Neural Networks for Space Sensor Applications

A smart-pixel cellular neural network (CNN) with hardware annealing capability, digitally programmable synaptic weights, and multisensor parallel interface has been under development for advanced space sensor applications. The smart-pixel CNN architecture is a programmable multi-dimensional array of optoelectronic neurons which are locally connected with their local neurons and associated active-pixel sensors. Integration of the neuroprocessor in each processor node of a scalable multiprocessor system offers orders-of-magnitude computing performance enhancements for on-board real-time intelligent multisensor processing and control tasks of advanced small satellites. The smart-pixel CNN operation theory, architecture, design and implementation, and system applications are investigated in detail. The VLSI (Very Large Scale Integration) implementation feasibility was illustrated by a prototype smart-pixel 5x5 neuroprocessor array chip of active dimensions 1380 micron x 746 micron in a 2-micron CMOS technology.

Fang, Wai-Chi↗
NASA NTRS2001

Accuracy Assessment of Response Surface Approximations for Supersonic Turbine Design

There is a growing trend to employ CFD tools to supply the necessary information for design optimization of fluid dynamics components/systems. Such results are prone to uncertainties due to reasons including discretization. errors, incomplete convergence of computational procedures, and errors associated with physical models such as turbulence closures. Based on this type of information, gradient-based optimization algorithms often suffer from the noisy calculations, which can seriously compromise the outcome. Similar problems arise from the experimental measurements. Global optimization techniques, such as those based on the response surface (RS) concept are becoming popular in part because they can overcome some of these barriers. However, there are also fundamental issues related to such global optimization technique such as RS. For example, in high dimensional design spaces, typically only a small number of function evaluations are available due to computational and experimental costs. On the other hand, complex features of the design variables do not allow one to model the global characteristics of the design space with simple quadratic polynomials. Consequently a main challenge is to reduce the size of the region where we fit the RS, or make it more accurate in the regions where the optimum is likely to reside. Response Surface techniques using either polynomials or and Neural Network (NN) methods offer designers alternatives to conduct design optimization. The RS technique employs statistical and numerical techniques to establish the relationship between design variables and objective/constraint functions, typically using polynomials. In this study, we aim at addressing issues related to the following questions: (1) How to identify outliers associated with a given RS representation and improve the RS model via appropriate treatments? (2) How to focus on selected design data so that RS can give better performance in regions critical to design optimization? (3) How to combine NN and polynomial techniques for improving the accuracy of the RS model?

Papila, Nilay↗
NASA NTRS1997

Incongruity, Incongruity Resolution, and Mental States: The Measure and Modification of Situational Awareness and Control

Cognition and emotion combine to define mental states. Situational awareness depends on both knowledge of the environment and the mood of the individual. Cognitive scientists from William James and Sigmond Freud to contemporary theorists in artificial intelligence and neuropsychology have acknowledged the critical role of subjective state in determining the efficiency and flexibility of information processing. One of the most explicit computational models of mental states to incorporate both knowledge and arousal has been described. Knowledge is carried in a typical neural net with categorical nodes and probabilistic links. Arousal determines the focus among these nodes and links. High arousal results in a restricted range of activation. Low arousal causes a wider range of stimulation and a broader linking of categories or "ideas." From this model Gerlernter generates "creativity" in problem solving from a network that is widely active and the possibility of "fixation" from a highly aroused system.

Derks, Peter L.↗
NASA NTRS1990

Photonic processing at NASA Ames Research Center

The Photonic Processing group is engaged in applied research on optical processors in support of the Ames vision to lead the development of autonomous intelligent systems. Optical processors, in conjunction with numeric and symbolic processors, are needed to provide the powerful processing capability that is required for many future agency missions. The research program emphasizes application of analog optical processing, where free-space propagation between components allows natural implementations of algorithms requiring a large degree of parallel computation. Special consideration is given in the Ames program to the integration of optical processors into larger, heterogeneous computational systems. Demonstration of the effective integration of optical processors within a broader knowledge-based system is essential to evaluate their potential for dependable operation in an autonomous environment such as space. The Ames Photonics program is currently addressing several areas of interest. One of the efforts is to develop an optical correlator system with two programmable spatial light modulators (SLMs) to perform distortion invariant pattern recognition. Another area of research is optical neural networks, also for use in distortion-invariant pattern recognition.

Ochoa, Ellen↗
NASA NTRSDate not supplied

Applying the Cognitive Space Gateway to Swarm Topologies

NASA’s future vision for interplanetary networking includes a lunar network, Cube Satellite (CubeSat) constellations, and deep space robotic missions, comprising what could be viewed as a network of networks. Delay-tolerant networking (DTN) architecture and protocols provide a standard network layer among these varying scenarios and mitigate many challenges of the space environment, such as long delays, unplanned service interruptions, and asymmetric links. The Cognitive Space Gateway (CSG) is a routing method in a DTN architecture that uses spiking neural networks as the learning element to optimize outing decisions in a complex environment. This work aims to further develop cognitive networking technologies in several critical areas, including DTN, the CSG algorithm, SmallSat swarm topologies, and cloud services. The CSG algorithm is tested in a realistic scenario in which the emulated network topology is based on a SmallSat swarm. The emulation environment will be built upon a commercial cloud service, such as Amazon Web Services (AWS) Elastic Compute Cloud. This work investigates the ability of such a platform to enable a flexible, lower maintenance approach to creating a multihop network outside of a physical laboratory. The cloud platform will provide a secure environment allowing for collaboration among government and academic entities.

Ricardo Lent↗
NASA NTRS1993

Visual information processing II; Proceedings of the Meeting, Orlando, FL, Apr. 14-16, 1993

Various papers on visual information processing are presented. Individual topics addressed include: aliasing as noise, satellite image processing using a hammering neural network, edge-detetion method using visual perception, adaptive vector median filters, design of a reading test for low-vision image warping, spatial transformation architectures, automatic image-enhancement method, redundancy reduction in image coding, lossless gray-scale image compression by predictive GDF, information efficiency in visual communication, optimizing JPEG quantization matrices for different applications, use of forward error correction to maintain image fidelity, effect of peanoscanning on image compression. Also discussed are: computer vision for autonomous robotics in space, optical processor for zero-crossing edge detection, fractal-based image edge detection, simulation of the neon spreading effect by bandpass filtering, wavelet transform (WT) on parallel SIMD architectures, nonseparable 2D wavelet image representation, adaptive image halftoning based on WT, wavelet analysis of global warming, use of the WT for signal detection, perfect reconstruction two-channel rational filter banks, N-wavelet coding for pattern classification, simulation of image of natural objects, number-theoretic coding for iconic systems.

Huck, Friedrich O.↗
NASA NTRS1989

Cloud field classification based on textural features

An essential component in global climate research is accurate cloud cover and type determination. Of the two approaches to texture-based classification (statistical and textural), only the former is effective in the classification of natural scenes such as land, ocean, and atmosphere. In the statistical approach that was adopted, parameters characterizing the stochastic properties of the spatial distribution of grey levels in an image are estimated and then used as features for cloud classification. Two types of textural measures were used. One is based on the distribution of the grey level difference vector (GLDV), and the other on a set of textural features derived from the MaxMin cooccurrence matrix (MMCM). The GLDV method looks at the difference D of grey levels at pixels separated by a horizontal distance d and computes several statistics based on this distribution. These are then used as features in subsequent classification. The MaxMin tectural features on the other hand are based on the MMCM, a matrix whose (I,J)th entry give the relative frequency of occurrences of the grey level pair (I,J) that are consecutive and thresholded local extremes separated by a given pixel distance d. Textural measures are then computed based on this matrix in much the same manner as is done in texture computation using the grey level cooccurrence matrix. The database consists of 37 cloud field scenes from LANDSAT imagery using a near IR visible channel. The classification algorithm used is the well known Stepwise Discriminant Analysis. The overall accuracy was estimated by the percentage or correct classifications in each case. It turns out that both types of classifiers, at their best combination of features, and at any given spatial resolution give approximately the same classification accuracy. A neural network based classifier with a feed forward architecture and a back propagation training algorithm is used to increase the classification accuracy, using these two classes of features. Preliminary results based on the GLDV textural features alone look promising.

Sengupta, Sailes Kumar↗
NASA NTRS1992

Implementation of an intelligent control system

A laboratory testbed facility which was constructed at NASA LeRC for the development of an Intelligent Control System (ICS) for reusable rocket engines is described. The framework of the ICS consists of a hierarchy of various control and diagnostic functions. The traditional high speed, closed-loop controller resides at the lowest level of the ICS hierarchy. Above this level resides the diagnostic functions which identify engine faults. The ICS top level consists of the coordination function which manages the interaction between an expert system and a traditional control system. The purpose of the testbed is to demonstrate the feasibility of the OCS concept by implementing the ICS as the primary controller in a simulation of the Space Shuttle Main Engine (SSME). The functions of the ICS which are implemented in the testbed are as follows: an SSME dynamic simulation with selected fault mode models, a reconfigurable controller, a neural network for sensor validation, a model-based failure detection algorithm, a rule based failure detection algorithm, a diagnostic expert system, an intelligent coordinator, and a user interface which provides a graphical representation of the event occurring within the testbed. The diverse nature of the ICS has led to the development of a distributed architecture consisting of specialized hardware and software for the implementation of the various functions. This testbed is made up of five different computer systems. These individual computers are discussed along with the schemes used to implement the various ICS components. The communication between computers and the timing and synchronization between components are also addressed.

Simon, D. L.↗
NASA NTRS2005

Integrated System-Level Optimization for Concurrent Engineering With Parametric Subsystem Modeling

The introduction of concurrent design practices to the aerospace industry has greatly increased the productivity of engineers and teams during design sessions as demonstrated by JPL's Team X. Simultaneously, advances in computing power have given rise to a host of potent numerical optimization methods capable of solving complex multidisciplinary optimization problems containing hundreds of variables, constraints, and governing equations. Unfortunately, such methods are tedious to set up and require significant amounts of time and processor power to execute, thus making them unsuitable for rapid concurrent engineering use. This paper proposes a framework for Integration of System-Level Optimization with Concurrent Engineering (ISLOCE). It uses parametric neural-network approximations of the subsystem models. These approximations are then linked to a system-level optimizer that is capable of reaching a solution quickly due to the reduced complexity of the approximations. The integration structure is described in detail and applied to the multiobjective design of a simplified Space Shuttle external fuel tank model. Further, a comparison is made between the new framework and traditional concurrent engineering (without system optimization) through an experimental trial with two groups of engineers. Each method is evaluated in terms of optimizer accuracy, time to solution, and ease of use. The results suggest that system-level optimization, running as a background process during integrated concurrent engineering sessions, is potentially advantageous as long as it is judiciously implemented.

Schuman, Todd↗