Search NASA⌕ Search

SEARCH · Search NASA

Results for “Loss function”

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 91 records · Page 5
All sourcesNASA NTRSDOE OSTI
NASA NTRS2019

An Ensemble of Bayesian Neural Networks for Exoplanetary Atmospheric Retrieval

Machine learning (ML) is now used in many areas of astrophysics, from detecting exoplanets in Kepler transit signals to removing telescope systematics. Recent work demonstrated the potential of using ML algorithms for atmospheric retrieval by implementing a random forest (RF) to perform retrievals in seconds that are consistent with the traditional, computationally expensive nested-sampling retrieval method. We expand upon their approach by presenting a new ML model, plan-net, based on an ensemble of Bayesian neural networks (BNNs) that yields more accurate inferences than the RF for the same data set of synthetic transmission spectra. We demonstrate that an ensemble provides greater accuracy and more robust uncertainties than a single model. In addition to being the first to use BNNs for atmospheric retrieval, we also introduce a new loss function for BNNs that learns correlations between the model outputs. Importantly, we show that designing ML models to explicitly incorporate domain-specific knowledge both improves performance and provides additional insight by inferring the covariance of the retrieved atmospheric parameters. We apply plan-net to the Hubble Space Telescope Wide Field Camera 3 transmission spectrum for WASP-12b and retrieve an isothermal temperature and water abundance consistent with the literature. We highlight that our method is flexible and can be expanded to higher resolution spectra and a larger number of atmospheric parameters.

Adam D. Cobb↗
NASA NTRSDate not supplied

Anomaly Detection in Flight Operational Data Using Deep Learning

In this session, we demonstrate two recently developed deep learning models for anomaly detection in flight operational data by the Data Sciences Group at NASA Ames Research Center. The first model is Convolutional Variational Auto-Encoder (CVAE) [1], which is an unsupervised deep encoder-decoder model, designed specifically for finding anomalies in heterogeneous multivariate time series data. We will demonstrate its application to finding anomalies in streaming data from NASA’s Digital Information Platform’s Fuser source. CVAE identifies data instances that are not representative of expected nominal behavior as anomalous. Since it is an unsupervised approach, the flagged anomalies will need to be reviewed by the subject matter experts (SMEs) for validation and labeling and is designed to assist with vulnerability discovery within Safety Monitoring System programs. The second model is Robust and Explainable Semi-supervised Anomaly Detection (RESAD) model [2], which builds on CVAE to allow learning from both minimally labeled data (previously reviewed by the SMEs) as well as majority unlabeled data. RESAD takes advantage of graph theoretic techniques to propagate the labels from the labeled data to the unlabeled data based on a pre-defined similarity metric and structures the learned feature space from flight time-series so that data of the same class would cluster tightly together. This model characteristic is enabled by training with an augmented loss function and allows learning of a more informative feature space for down-stream tasks such as search and active learning. We demonstrate RESAD using data from the NASA DASHlink project [3].

anomaly detection↗
NASA NTRSDate not supplied

Differential Equation Approximation Using Gradient-Boosted Quantile Regression

The operation of cyber-physical-human (CPH) systems is subject to various epistemic and aleatory uncertainties. Overall trustworthiness of CPH systems relies on the trustworthiness of its components and their interactions. It is important that computational models comprising the cyber component of CPH provide predictions accompanied by a measure of confidence in model outcomes. Uncertainty quantification (UQ) and propagation are especially important in safety critical CPH systems. Gradient-boosted trees is a modeling approach capable both of learning the dynamics of a system and performing UQ. In this paper, we devise a method for using gradient boosting to learn the dynamics of a second order differential equation and estimate uncertainty at the same time. We do this by creating a custom loss function that trains the model to approximate the second derivative of a noisy time series, and to penalize based on a parameter that corresponds to the desired quantile. The resulting gradient boosting model can simulate stochastic trajectories of the system given a single starting point, that is, it can estimate both the expected trajectory and its uncertainty. We show that the uncertainty estimation is well calibrated and that the model can learn the dynamics even in the presence of noise. We demonstrate the approach on a simple cartpole system.

Autonomous systems↗
NASA NTRS2022

Deep Interacting Multiple Model Filtering

In this paper, a deep learning-based multiple model estimation framework is presented for the state estimation of hybrid dynamical systems from high dimensional observations such as camera images. A low dimensional vector which represents the measurement of the latent dynamical system and its corresponding variance are learned using a deep encoder neural network. An Interacting Multiple Model (IMM) filter is used to generate the latent state estimates and covariances using multiple dynamical models, which can be learned using backpropagation through time. The state estimates of the dynamical system and the corresponding covariance matrix are generated from the latent state estimates and covariance using a deep decoder neural network. The whole network is trained in an end-to-end manner using a loss function which minimizes the negative log-likelihood of the neural network parameters. Simulation results are presented using a 2D bouncing ball example and estimation error statistics are computed which demonstrates the accuracy and consistency of the estimation.

Ghananeel Rotithor↗
NASA NTRSDate not supplied

A Neural Network Parametrization of Volumetric Cloud Fraction Profiles Using Satellite Observations and MERRA-2 Reanalysis Meteorological Data

Clouds play a crucial role in regulating the hydrologic cycle and Earth's radiative energy budget, yet they are often poorly represented in global climate models (GCMs). This study applies deep machine learning techniques to develop a physical parameterization of volumetric cloud fraction (VCF), the fraction of a 3-D grid volume occupied by clouds using satellite lidar-radar measurements. The neural network (NN) captures the complicated relationships between observed VCF profiles and collocated meteorological variables from MERRA-2 reanalysis data. Our results show that the NN model, particularly a sequence-to-sequence long short-term memory (LSTM) network with a sixfactor loss function, effectively learns the underlying cloud physical processes. The NN model outperforms MERRA-2 reanalysis in representing low-level clouds in tropical and subtropical regions and low- and middle-level clouds over midlatitude storm-track regions, and also improves VCF histograms. These improvements are reflected in the vertical distributions of zonally, meridionally, and globally averaged VCFs, geographic distributions of low-, middle-, and high-level clouds, and seasonal variations in monthly-mean VCF. Furthermore, the NN predictions effectively capture the El Niño-Southern Oscillation (ENSO) effects and other interannual variations. The NN parameterization is further evaluated through a sensitivity analysis, in which a single predictor is perturbed at a time. This reveals that relative humidity (RH) is the dominant factor influencing variations in globally averaged VCF at low and middle altitudes, followed by temperature. At higher altitudes, temperature becomes the primary driver of VCF through its effect on RH. Changes in wind components had minimal impact on globally averaged VCF.

Shan Zeng↗
NASA NTRS2007

Recovery of Spaceflight-induced Bone Loss: Bone Mineral Density after Long-Duration Missions as Fitted with an Exponential Function

The loss of bone mineral in NASA astronauts during spaceflight has been investigated throughout the more than 40 years of space travel. Consequently, it is a medical requirement at NASA Johnson Space Center (JSC) that changes in bone mass be monitored in crew members by measuring bone mineral density (BMD) with dual-energy x-ray absorptiometry (DXA) before and after flight on astronauts who serve on long-duration missions (4-6 months). We evaluated this repository of medical data to track whether there is recovery of bone mineral that was lost during spaceflight. Our analysis was supplemented by BMD data from cosmonauts ( by convention, a space traveler formally employed by the Russia Aviation and Space Agency or by the previous Soviet Union) who had also flown on long-duration missions. Data from a total of 45 individual crew members -- a small number of whom flew on more than one mission -- were used in this analysis. Changes in BMD (between 56 different sets of pre- and postflight measurements) were plotted as a function of time (days after landing). Plotted BMD changes were fitted to an exponential mathematical function that estimated: i) BMD change on landing day (day 0) and ii) the number of days after landing when 50% of the lost bone would be recovered ("50% recovery time") in the lumbar spine, trochanter, pelvis, femoral neck and calcaneus. In sum, averaged losses of bone mineral after long-duration spaceflight ranged between 2-9% across all sites with our recovery model predicting a 50% restoration of bone loss for all sites to be within 9 months.

Sibonga, J. D.↗
NASA NTRS2003

Biochemical and functional analysis of CTR1, a protein kinase that negatively regulates ethylene signaling in Arabidopsis

CTR1 encodes a negative regulator of the ethylene response pathway in Arabidopsis thaliana. The C-terminal domain of CTR1 is similar to the Raf family of protein kinases, but its first two-thirds encodes a novel protein domain. We used a variety of approaches to investigate the function of these two CTR1 domains. Recombinant CTR1 protein was purified from a baculoviral expression system, and shown to possess intrinsic Ser/Thr protein kinase activity with enzymatic properties similar to Raf-1. Deletion of the N-terminal domain did not elevate the kinase activity of CTR1, indicating that, at least in vitro, this domain does not autoinhibit kinase function. Molecular analysis of loss-of-function ctr1 alleles indicated that several mutations disrupt the kinase catalytic domain, and in vitro studies confirmed that at least one of these eliminates kinase activity, which indicates that kinase activity is required for CTR1 function. One missense mutation, ctr1-8, was found to result from an amino acid substitution within a new conserved motif within the N-terminal domain. Ctr1-8 has no detectable effect on the kinase activity of CTR1 in vitro, but rather disrupts the interaction with the ethylene receptor ETR1. This mutation also disrupts the dominant negative effect that results from overexpression of the CTR1 amino-terminal domain in transgenic Arabidopsis. These results suggest that CTR1 interacts with ETR1 in vivo, and that this association is required to turn off the ethylene-signaling pathway.

Non-NASA Center↗
NASA NTRS1999

Molecular Signaling in Muscle Plasticity

Extended spaceflight under microgravity conditions leads to significant atrophy of weight-bearing muscles. Atrophy and hypertrophy are the extreme outcomes of the high degree of plasticity exhibited by skeletal muscle. Stimuli which control muscle plasticity include neuronal, hormonal, nutritional, and mechanical inputs. The mechanical stimulus for muscle is directly related to the work or exercise against a load performed. Little or no work is performed by weight-bearing muscles under microgravity conditions. A major hypothesis is that focal adhesion kinase (FAK) which is associated with integrin at the adherens junctions and costa meres of all skeletal muscles is an integral part of the major mechanism for molecular signaling upon mechanical stimulation in all muscle fibers. Additionally, we propose that myotonic protein kinase (DMPK) and dystrophin (DYSTR) also participate in distinct mechanically stimulated molecular signaling pathways that are most critical in type I and type II muscle fibers, respectively. To test these hypotheses, we will use the paradigms of hindlimb unloading and overloading in mice as models for microgravity conditions and a potential exercise countermeasure, respectively, in mice. We expect that FAK loss-of-function will impair hypertrophy and enhance atrophy in all skeletal muscle fibers whereas DYSTR and DMPK loss-of-function will have similar but more selective effects on Type IT and Type I fibers, respectively. Gene expression will be monitored by muscle-specific creatine kinase M promoter-reporter construct activity and specific MRNA and protein accumulation in the soleus (type I primarily) and plantaris (type 11 primarily) muscles. With these paradigms and assays, the following Specific Project Aims will be tested in genetically altered mice: 1) identify the roles of DYSTR and its pathway; 2) evaluate the roles of the DMPK and its pathway; 3) characterize the roles of FAK and its pathway and 4) genetically analyze the mechanisms and interactions between the FAK, DYSTR, and DMPK-associated pathways in single and specific combinations of mutants. The identification of potential signaling mechanisms may permit future development of pharmacological countermeasures for amelioration and prevention of the microgravity-induced atrophy in extended spaceflight, and the analysis of both overloading and unloading paradigms may provide further support for development of exercise-based countermeasures. Understanding the basic mechanisms of molecular signaling in muscle plasticity may aid our understanding and treatment of skeletal muscle atrophy not only in spaceflight but in similar problems of the aging population, in prolonged bed rest, and in cachexia associated with chronic disease.

Epstein, Henry F.↗
NASA NTRS2002

Alterations in protein metabolism during space flight and inactivity

Space flight and the accompanying diminished muscular activity lead to a loss of body nitrogen and muscle function. These losses may affect crew capabilities and health in long-duration missions. Space flight alters protein metabolism such that the body is unable to maintain protein synthetic rates. A concomitant hypocaloric intake and altered anabolic/catabolic hormonal profiles may contribute to or exacerbate this problem. The inactivity associated with bedrest also reduces muscle and whole-body protein synthesis. For this reason, bedrest provides a good model for the investigation of potential exercise and nutritional countermeasures to restore muscle protein synthesis. We have demonstrated that minimal resistance exercise preserves muscle protein synthesis throughout bedrest. In addition, ongoing work indicates that an essential amino acid and carbohydrate supplement may ameliorate the loss of lean body mass and muscle strength associated with 28 d of bedrest. The investigation of inactivity-induced alterations in protein metabolism, during space flight or prolonged bedrest, is applicable to clinical populations and, in a more general sense, to the problems associated with the decreased activity that occur with aging.

Non-NASA Center↗
NASA NTRS2018

Cumulative Distribution of Ballistic Impact Failures of Common Twisted-Pair Data Cables at Orbital Speeds

Data wire cable runs are a significant presence on the exterior of the International Space Station (ISS), and continued ISS mission support requires detailed assessment of cables due to micrometeoroid and orbit debris (MMOD) impact. These data wire cables are twisted-pair cables consisting of two 22AWG stranded conductors and fillers inside a tight fitting braided copper shield. The copper shield and its contents are covered with a jacket that has a nominal outer diameter of 3.76 mm and beta-cloth tape. The ISS engineering community has identified two loss-of-function mechanisms for these cables: open circuits due to severed conductors within the cable, and short circuits due to contact between conductors or grounded components. As these data cables are low power systems, short circuits are not expected to burn away the contact, so both open and short circuits are considered permanent loss-of-function for the cable. A total of ninety-seven impact experiments have been performed into these cables to develop a statistical model for the failure of these cables to be used in reliability studies. The experimental work has yielded cumulative distribution functions for these cables for steel and aluminum components of the orbital debris environment at representative speeds and impact obliquities.

Miller, Joshua E.↗
NASA NTRS2013

Piezoelectrically Initiated Pyrotechnic Igniter

This innovation consists of a pyrotechnic initiator and piezoelectric initiation system. The device will be capable of being initiated mechanically; resisting initiation by EMF, RF, and EMI (electromagnetic field, radio frequency, and electromagnetic interference, respectively); and initiating in water environments and space environments. Current devices of this nature are initiated by the mechanical action of a firing pin against a primer. Primers historically are prone to failure. These failures are commonly known as misfires or hang-fires. In many cases, the primer shows the dent where the firing pin struck the primer, but the primer failed to fire. In devices such as "T" handles, which are commonly used to initiate the blowout of canopies, loss of function of the device may result in loss of crew. In devices such as flares or smoke generators, failure can result in failure to spot a downed pilot. The piezoelectrically initiated ignition system consists of a pyrotechnic device that plugs into a mechanical system (activator), which on activation, generates a high-voltage spark. The activator, when released, will strike a stack of electrically linked piezo crystals, generating a high-voltage, low-amperage current that is then conducted to the pyro-initiator. Within the initiator, an electrode releases a spark that passes through a pyrotechnic first-fire mixture, causing it to combust. The combustion of the first-fire initiates a primary pyrotechnic or explosive powder. If used in a "T" handle, the primary would ramp the speed of burn up to the speed of sound, generating a shock wave that would cause a high explosive to go "high order." In a flare or smoke generator, the secondary would produce the heat necessary to ignite the pyrotechnic mixture. The piezo activator subsystem is redundant in that a second stack of crystals would be struck at the same time with the same activation force, doubling the probability of a first strike spark generation. If the first activation fails to ignite, the device is capable of multiple attempts. Another unique aspect is in the design of the pyrotechnic device. There is an electrode that aids the generation of a directed spark and the use of a conductive matrix to support the first-fire material so that the spark will penetrate to the second electrode.

Quince, Asia↗
NASA NTRS2003

Evaporation Loss of Light Elements as a Function of Cooling Rate: Logarithmic Law

Knowledge about the evaporation loss of light elements is important to our understanding of chondrule formation processes. The evaporative loss of light elements (such as B and Li) as a function of cooling rate is of special interest because recent investigations of the distribution of Li, Be and B in meteoritic chondrules have revealed that Li varies by 25 times, and B and Be varies by about 10 times. Therefore, if we can extrapolate and interpolate with confidence the evaporation loss of B and Li (and other light elements such as K, Na) at a wide range of cooling rates of interest based upon limited experimental data, we would be able to assess the full range of scenarios relating to chondrule formation processes. Here, we propose that evaporation loss of light elements as a function of cooling rate should obey the logarithmic law.

Xiong, Yong-Liang↗
NASA NTRS2019

Update: Antarctic Winterover as an Analog for Spaceflight Immune Dysregulation

Orbital spaceflight perturbs the human immune system significantly; Natural Killer (NK) and T-lymphocyte (T) cell functions are most susceptible to spaceflight-induced impairment. This loss of function may manifest in persistent latent virus reactivation (CMV, EBV, VZV), which does occur at a higher frequency in astronauts compared to earthlings.

Makedonas, George↗
NASA NTRS1999

Activation of the Saccharomyces cerevisiae filamentation/invasion pathway by osmotic stress in high-osmolarity glycogen pathway mutants

Mitogen-activated protein kinase (MAPK) cascades are frequently used signal transduction mechanisms in eukaryotes. Of the five MAPK cascades in Saccharomyces cerevisiae, the high-osmolarity glycerol response (HOG) pathway functions to sense and respond to hypertonic stress. We utilized a partial loss-of-function mutant in the HOG pathway, pbs2-3, in a high-copy suppressor screen to identify proteins that modulate growth on high-osmolarity media. Three high-copy suppressors of pbs2-3 osmosensitivity were identified: MSG5, CAK1, and TRX1. Msg5p is a dual-specificity phosphatase that was previously demonstrated to dephosphorylate MAPKs in yeast. Deletions of the putative MAPK targets of Msg5p revealed that kss1delta could suppress the osmosensitivity of pbs2-3. Kss1p is phosphorylated in response to hyperosmotic shock in a pbs2-3 strain, but not in a wild-type strain nor in a pbs2-3 strain overexpressing MSG5. Both TEC1 and FRE::lacZ expressions are activated in strains lacking a functional HOG pathway during osmotic stress in a filamentation/invasion-pathway-dependent manner. Additionally, the cellular projections formed by a pbs2-3 mutant on high osmolarity are absent in strains lacking KSS1 or STE7. These data suggest that the loss of filamentation/invasion pathway repression contributes to the HOG mutant phenotype.

Non-NASA Center↗
NASA NTRS2005

Electroporation System for Sterilizing Water

A prototype of an electroporation system for sterilizing wastewater or drinking water has been developed. In electroporation, applied electric fields cause transient and/or permanent changes in the porosities of living cells. Electroporation at lower field strengths can be exploited to increase the efficiency of chemical disinfection (as in chlorination). Electroporation at higher field strengths is capable of inactivating and even killing bacteria and other pathogens, without use of chemicals. Hence, electroporation is at least a partial alternative to chlorination. The transient changes that occur in micro-organisms at lower electric-field strengths include significantly increased uptake of ions and molecules. Such increased uptake makes it possible to achieve disinfection at lower doses of chemicals (e.g., chlorine or ozone) than would otherwise be needed. Lower doses translate to lower costs and reduced concentrations of such carcinogenic chemical byproducts as trichloromethane. Higher electric fields cause cell membranes to lose semipermeability and thereby become unable to function as selective osmotic barriers between the cells and the environment. This loss of function is the cause of the cell death at higher electric-field intensities. Experimental evidence does not indicate cell lysis but, rather, combined leaking of cell proteins out of the cells as well as invasion of foreign chemical compounds into the cells. The concept of electroporation is not new: it has been applied in molecular biology and genetic engineering for decades. However, the laboratory-scale electroporators used heretofore have been built around small (400-microliter) cuvettes, partly because the smallness facilitates the generation of electric fields of sufficient magnitude to cause electroporation. Moreover, most laboratory- scale electroporators have been designed for testing static water. In contrast, the treatment cell in the present system is much larger and features a flow-through geometry, such that electric fields strong enough to effect 99.9- percent disinfection can be applied to water flowing in a pipe.

Schlager, Kenneth J.↗
NASA NTRS2011

Reliability Impacts in Life Support Architecture and Technology Selection

Equivalent System Mass (ESM) and reliability estimates were performed for different life support architectures based primarily on International Space Station (ISS) technologies. The analysis was applied to a hypothetical 1-year deep-space mission. High-level fault trees were initially developed relating loss of life support functionality to the Loss of Crew (LOC) top event. System reliability was then expressed as the complement (nonoccurrence) this event and was increased through the addition of redundancy and spares, which added to the ESM. The reliability analysis assumed constant failure rates and used current projected values of the Mean Time Between Failures (MTBF) from an ISS database where available. Results were obtained showing the dependence of ESM on system reliability for each architecture. Although the analysis employed numerous simplifications and many of the input parameters are considered to have high uncertainty, the results strongly suggest that achieving necessary reliabilities for deep-space missions will add substantially to the life support system mass. As a point of reference, the reliability for a single-string architecture using the most regenerative combination of ISS technologies without unscheduled replacement spares was estimated to be less than 1%. The results also demonstrate how adding technologies in a serial manner to increase system closure forces the reliability of other life support technologies to increase in order to meet the system reliability requirement. This increase in reliability results in increased mass for multiple technologies through the need for additional spares. Alternative parallel architecture approaches and approaches with the potential to do more with less are discussed. The tall poles in life support ESM are also reexamined in light of estimated reliability impacts.

Lange, Kevin E.↗