Search NASA⌕ Search

SEARCH · Search NASA

Results for “evolutionary computation”

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 451 records · Page 25

Application of an Evolutionary Algorithm for Parameter Optimization in a Gully Erosion Model

Herein we demonstrate how to use model optimization to determine a set of best-fit parameters for a landform model simulating gully incision and headcut retreat. To achieve this result we employed the Covariance Matrix Adaptation Evolution Strategy (CMA-ES), an iterative process in which samples are created based on a distribution of parameter values that evolve over time to better fit an objective function. CMA-ES efficiently finds optimal parameters, even with high-dimensional objective functions that are non-convex, multimodal, and non-separable. We ran model instances in parallel on a high-performance cluster, and from hundreds of model runs we obtained the best parameter choices. This method is far superior to brute-force search algorithms, and has great potential for many applications in earth science modeling. We found that parameters representing boundary conditions tended to converge toward an optimal single value, whereas parameters controlling geomorphic processes are defined by a range of optimal values.

97 MATHEMATICS AND COMPUTING↗

Naturally ornate RNA-only complexes revealed by cryo-EM

The structures of natural RNAs remain poorly characterized and may hold numerous surprises. Here we report three-dimensional structures of three large ornate bacterial RNAs using cryo-electron microscopy (cryo-EM). GOLLD (Giant, Ornate, Lake- and Lactobacillales-Derived), ROOL (Rumen-Originating, Ornate, Large) and OLE (Ornate Large Extremophilic) RNAs form homo-oligomeric complexes whose stoichiometries are retained at lower concentrations than measured in cells. OLE RNA forms a dimeric complex with long co-axial pipes spanning two monomers. Both GOLLD and ROOL form distinct RNA-only multimeric nanocages with diameters larger than the ribosome, each empty except for a disordered loop. Extensive intramolecular and intermolecular A-minor interactions, kissing loops, an unusual A–A helix and other interactions stabilize the three complexes. Sequence covariation analysis of these large RNAs reveals evolutionary conservation of intermolecular interactions, supporting the biological importance of large, ornate RNA quaternary structures that can assemble without any involvement of proteins.

59 BASIC BIOLOGICAL SCIENCES↗

Single-cell genomics-based analysis of virus–host interactions in marine surface bacterioplankton

Viral infections dynamically alter the composition and metabolic potential of marine microbial communities and the evolutionary trajectories of host populations with resulting feedback on biogeochemical cycles. It is quite possible that all microbial populations in the ocean are impacted by viral infections. Our knowledge of virus–host relationships, however, has been limited to a minute fraction of cultivated host groups. Here, we utilized single-cell sequencing to obtain genomic blueprints of viruses inside or attached to individual bacterial and archaeal cells captured in their native environment, circumventing the need for host and virus cultivation. Furthermore, a combination of comparative genomics, metagenomic fragment recruitment, sequence anomalies and irregularities in sequence coverage depth and genome recovery were utilized to detect viruses and to decipher modes of virus–host interactions. Members of all three tailed phage families were identified in 20 out of 58 phylogenetically and geographically diverse single amplified genomes (SAGs) of marine bacteria and archaea. At least four phage–host interactions had the characteristics of late lytic infections, all of which were found in metabolically active cells. One virus had genetic potential for lysogeny. Our findings include first known viruses of Thaumarchaeota, Marinimicrobia, Verrucomicrobia and Gammaproteobacteria clusters SAR86 and SAR92. Viruses were also found in SAGs of Alphaproteobacteria and Bacteroidetes. A high fragment recruitment of viral metagenomic reads confirmed that most of the SAG-associated viruses are abundant in the ocean. This study demonstrates that single-cell genomics, in conjunction with sequence-based computational tools, enable in situ, cultivation-independent insights into host–virus interactions in complex microbial communities.

58 GEOSCIENCES↗

Toward Consistent High-Fidelity Quantum Learning on Unstable Devices via Efficient In-Situ Calibration

In the near-term noisy intermediate-scale quantum (NISQ) era, high noise will significantly reduce the fidelity of quantum computing. What's worse, recent works reveal that the noise on quantum devices is not stable, that is, the noise is dynamically changing over time. This leads to an imminent challenging problem: At run-time, is there a way to efficiently achieve a consistent high-fidelity quantum system on unstable devices? To study this problem, we take quantum learning (a.k.a., variational quantum algorithm) as a vehicle, which has a wide range of applications, such as combinatorial optimization and machine learning. A straightforward approach is to optimize a variational quantum circuit (VQC) with a parameter-shift approach on the target quantum device before using it; however, the optimization has an extremely high time cost, which is not practical at run-time. To address the pressing issue, in this paper, we proposed a novel quantum pulse-based noise adaptation framework, namely QuPAD. In the proposed framework, first, we identify that the CNOT gate is the fidelity bottleneck of the conventional VQC, and we employ a more robust parameterized multi-qubit gate (i.e., Rzx gate) to replace CNOT gate. Second, by benchmarking Rzx gate with different parameters, we build a fitting function for each coupling qubit pair, such that the deviation between the theoretic output of Rzx gate and its on-device output under a given pulse amplitude and duration can be efficiently predicted. On top of this, an evolutionary algorithm is devised to identify the pulse amplitude and duration of each Rzx gate (i.e., calibration) and find the quantum circuits with high fidelity. Experiments show that the runtime on quantum devices of QuPAD with 8–10 qubits is less than 15 minutes, which is up to 270 x faster than the parameter-shift approach. In addition, compared to the vanilla VQC as a baseline, QuPAD can achieve 59.33% accuracy gain on a classification task, and average 66.34% closer to ground state energy for molecular simulation.

Hu, Zhirui↗

The Swan: Data-driven Inference of Stellar Surface Gravities for Cool Stars from Photometric Light Curves

Stellar light curves are well known to encode physical stellar properties. Precise, automated, and computationally inexpensive methods to derive physical parameters from light curves are needed to cope with the large influx of these data from space-based missions such as Kepler and TESS. Here we present a new methodology that we call “The Swan,” a fast, generalizable, and effective approach for deriving stellar surface gravity (logg) for main-sequence, subgiant, and red giant stars from Kepler light curves using local linear regression on the full frequency content of Kepler long-cadence power spectra. With this inexpensive data-driven approach, we recover logg to a precision of ~0.02 dex for 13,822 stars with seismic logg values between 0.2 and 4.4 dex and ~0.11 dex for 4646 stars with Gaia-derived logg values between 2.3 and 4.6 dex. We further develop a signal-to-noise metric and find that granulation is difficult to detect in many cool main-sequence stars (T {sub eff} ≲ 5500 K), in particular K dwarfs. By combining our logg measurements with Gaia radii, we derive empirical masses for 4646 subgiant and main-sequence stars with a median precision of ~7%. Finally, we demonstrate that our method can be used to recover logg to a similar mean absolute deviation precision for a TESS baseline of 27 days. Our methodology can be readily applied to photometric time series observations to infer stellar surface gravities to high precision across evolutionary states.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Stellar migration and chemical enrichment in the milky way disc: a hybrid model

ABSTRACT We develop a hybrid model of galactic chemical evolution that combines a multiring computation of chemical enrichment with a prescription for stellar migration and the vertical distribution of stellar populations informed by a cosmological hydrodynamic disc galaxy simulation. Our fiducial model adopts empirically motivated forms of the star formation law and star formation history, with a gradient in outflow mass loading tuned to reproduce the observed metallicity gradient. With this approach, the model reproduces many of the striking qualitative features of the Milky Way disc’s abundance structure: (i) the dependence of the [O/Fe]–[Fe/H] distribution on radius Rgal and mid-plane distance |z|; (ii) the changing shapes of the [O/H] and [Fe/H] distributions with Rgal and |z|; (iii) a broad distribution of [O/Fe] at sub-solar metallicity and changes in the [O/Fe] distribution with Rgal, |z|, and [Fe/H]; (iv) a tight correlation between [O/Fe] and stellar age for [O/Fe] > 0.1; (v) a population of young and intermediate-age α-enhanced stars caused by migration-induced variability in the Type Ia supernova rate; (vi) non-monotonic age–[O/H] and age–[Fe/H] relations, with large scatter and a median age of ∼4 Gyr near solar metallicity. Observationally motivated models with an enhanced star formation rate ∼2 Gyr ago improve agreement with the observed age–[Fe/H] and age–[O/H] relations, but worsen agreement with the observed age–[O/Fe] relation. None of our models predict an [O/Fe] distribution with the distinct bimodality seen in the observations, suggesting that more dramatic evolutionary pathways are required. All code and tables used for our models are publicly available through the Versatile Integrator for Chemical Evolution (VICE; https://pypi.org/project/vice).

79 ASTRONOMY AND ASTROPHYSICS↗

Troubleshooting deep-learner training data problems using an evolutionary algorithm on Summit

Architectural and hyper-parameter design choices can influence deep-learner (DL) model fidelity but can also be affected by malformed training and validation data. However, practitioners may spend significant time refining layers and hyper-parameters before discovering that distorted training data was impeding training progress. We found that an evolutionary algorithm (EA) can be used to diagnose this kind of DL problem. An EA evaluated thousands of DL configurations on Summit that yielded no overall improvement in DL performance, which suggested problems with the training and validation data. We suspected that Contrast Limited Adaptive Histogram Equalization (CLAHE) enhancement that was applied to previously generated digital surface models (DSMs), for which we were training DLs to find errors, had damaged the training data. Subsequent runs with an alternative global normalization yielded significantly improved DL performance. However, the DL Intersection Over Union (IOU) still exhibited consistent sub-par performance, which suggested further problems with the training data and DL approach. Nonetheless, we were able to diagnose this problem within a 12-hour span via Summit runs, which prevented several weeks of unproductive trial-and-error DL configuration refinement, and allowed for a more timely convergence on an ultimately viable solution.

97 MATHEMATICS AND COMPUTING↗

Binarity and the abundance discrepancy problem in planetary nebulae

The discrepancy between abundances computed using optical recombination lines and collisionally excited lines is a major unresolved problem in nebular astrophysics. Here, we show that the largest abundance discrepancies are reached in planetary nebulae with close binary central stars. We illustrate this using deep spectroscopy of three nebulae with a post common-envelope (CE) binary star. Abell 46 and Ou 5 have O{sup 2+}/H{sup +} abundance discrepancy factors larger than 50, and as high as 300 in the inner regions of Abell 46. Abell 63 has a smaller discrepancy factor around 10, which is still above the typical values in ionized nebulae. Our spectroscopic analysis supports previous conclusions that, in addition to “standard” hot (T{sub e} ∼ 10{sup 4} K) gas, there exists a colder (T{sub e} ∼ 10{sup 3} K), ionized component that is highly enriched in heavy elements. These nebulae have low ionized masses, between 10{sup −3} and 10{sup −1} M{sub ⊙} depending on the adopted electron densities and temperatures. Since the much more massive red giant envelope is expected to be entirely ejected in the CE phase, the currently observed nebulae would be produced much later, during post-CE mass loss episodes when the envelope has already dispersed. These observations add constraints to the abundance discrepancy problem. We revise possible explanations. Some explanations are naturally linked to binarity such as, for instance, high-metallicity nova ejecta, but it is difficult at this stage to depict an evolutionary scenario consistent with all of the observed properties. We also introduce the hypothesis that these nebulae are the result of tidal destruction, accretion, and ejection of Jupiter-like planets.

79 ASTRONOMY AND ASTROPHYSICS↗

VERY LARGE ARRAY AND ATCA SEARCH FOR NATAL STAR CLUSTERS IN NEARBY STAR-FORMING GALAXIES

In order to investigate the relationship between the local environment and the properties of natal star clusters, we obtained radio observations of 25 star-forming galaxies within 20 Mpc using the Very Large Array and the Australia Telescope Compact Array. Natal star-forming regions can be identified by their characteristic thermal radio emission, which is manifest in their spectral index at centimeter wavelengths. The host galaxies in our sample were selected based upon their likelihood of harboring young star formation. In star-forming regions, the ionizing flux of massive embedded stars powers the dominant thermal free-free emission of those sources, resulting in a spectral index of {alpha} {approx}> -0.2 (where S{sub {nu}} {proportional_to} {nu}{sup {alpha}}), which we compute. With the current sensitivity, we find that of the 25 galaxies in this sample only 5 have radio sources with spectral indices that are only consistent with a thermal origin, 4 have radio sources that are only consistent with a non-thermal origin, 6 have radio sources whose nature is ambiguous due to uncertainties in the spectral index, and 16 have no detected radio sources. For those sources that appear to be dominated by thermal emission, we infer the ionizing flux of the star clusters and the number of equivalent O7.5 V stars that are required to produce the observed radio flux densities. The most radio-luminous clusters that we detect have an equivalent of {approx}7 x 10{sup 3} O7.5 V stars, and the smallest only have an equivalent of {approx}10{sup 2} O7.5 V stars; thus these star-forming regions span the range of large OB associations to moderate 'super star clusters'. With the current detection limits, we also place upper limits on the masses of clusters that could have recently formed; for a number of galaxies we can conclusively rule out the presence of natal clusters significantly more massive than the Galactic star-forming region W49A ({approx}5 x 10{sup 4} M{sub sun}). The dearth of current massive cluster formation in these galaxies suggests that either their current star formation intensities have fallen to near or below that of the Milky Way and/or the evolutionary state that gives rise to thermal radio emission is short-lived.

79 ASTRONOMY AND ASTROPHYSICS↗

Geometry-Driven Detection, Tracking and Visual Analysis of Viscous and Gravitational Fingers

Viscous and gravitational flow instabilities cause a displacement front to break up into finger-like fluids. The detection and evolutionary analysis of these fingering instabilities are critical in multiple scientific disciplines such as fluid mechanics and hydrogeology. However, previous detection methods of the viscous and gravitational fingers are based on density thresholding, which provides limited geometric information of the fingers. The geometric structures of fingers and their evolution are important yet little studied in the literature. In this work, we explore the geometric detection and evolution of the fingers in detail to elucidate the dynamics of the instability. Additionally, we propose a ridge voxel detection method to guide the extraction of finger cores from three-dimensional (3D) scalar fields. After skeletonizing finger cores into skeletons, we design a spanning tree based approach to capture how fingers branch spatially from the finger skeletons. Finally, we devise a novel geometric-glyph augmented tracking graph to study how the fingers and their branches grow, merge, and split over time. Feedback from earth scientists demonstrates the usefulness of our approach to performing spatio-temporal geometric analyses of fingers.

97 MATHEMATICS AND COMPUTING↗

Origin and evolution of HIV-1 subtype A6

Background: HIV outbreaks in the Former Soviet Union (FSU) countries were characterized by repeated transmission of the HIV variant AFSU, which is now classified as a distinct subtype A sub-subtype called A6. The current study used phylogenetic/phylodynamic and signature mutation analyses to determine likely evolutionary relationship between subtype A6 and other subtype A sub-subtypes. Methods: For this study, an initial Maximum Likelihood phylogenetic analysis was performed using a total of 553 full-length, publicly available, reverse transcriptase sequences, from A1, A2, A3, A4, A5, and A6 sub-subtypes of subtype A. For phylogenetic clustering and signature mutation analysis, a total of 5961 and 3959 pol and env sequences, respectively, were used. Results: Phylogenetic and signature mutation analysis showed that HIV-1 sub-subtype A6 likely originated from sub-subtype A1 of African origin. A6 and A1 pol and env genes shared several signature mutations that indicate genetic similarity between the two subtypes. For A6, tMRCA dated to 1975, 15 years later than that of A1. Conclusion: The current study provides insights into the evolution and diversification of A6 in the backdrop of FSU countries and indicates that A6 in FSU countries evolved from A1 of African origin and is getting bridged outside the FSU region.

60 APPLIED LIFE SCIENCES↗

On the adaptivity and complexity embedded into differential evolution

This research deals with the comparison of the two modern approaches for evolutionary algorithms, which are the adaptivity and complex chaotic dynamics. This paper aims on the investigations on the chaos-driven Differential Evolution (DE) concept. This paper is aimed at the embedding of discrete dissipative chaotic systems in the form of chaotic pseudo random number generators for the DE and comparing the influence to the performance with the state of the art adaptive representative jDE. This research is focused mainly on the possible disadvantages and advantages of both compared approaches. Repeated simulations for Lozi map driving chaotic systems were performed on the simple benchmark functions set, which are more close to the real optimization problems. Obtained results are compared with the canonical not-chaotic and not adaptive DE. Results show that with used simple test functions, the performance of ChaosDE is better in the most cases than jDE and Canonical DE, furthermore due to the unique sequencing in CPRNG given by the hidden chaotic dynamics, thus better and faster selection of unique individuals from population, ChaosDE is faster.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Early prediction of antigenic transitions for influenza A/H3N2

Influenza A/H3N2 is a rapidly evolving virus which experiences major antigenic transitions every two to eight years. Anticipating the timing and outcome of transitions is critical to developing effective seasonal influenza vaccines. Using a published phylodynamic model of influenza transmission, we identified indicators of future evolutionary success for an emerging antigenic cluster and quantified fundamental trade-offs in our ability to make such predictions. The eventual fate of a new cluster depends on its initial epidemiological growth rate––which is a function of mutational load and population susceptibility to the cluster––along with the variance in growth rate across co-circulating viruses. Logistic regression can predict whether a cluster at 5% relative frequency will eventually succeed with ~80% sensitivity, providing up to eight months advance warning. As a cluster expands, the predictions improve while the lead-time for vaccine development and other interventions decreases. However, attempts to make comparable predictions from 12 years of empirical influenza surveillance data, which are far sparser and more coarse-grained, achieve only 56% sensitivity. By expanding influenza surveillance to obtain more granular estimates of the frequencies of and population-wide susceptibility to emerging viruses, we can better anticipate major antigenic transitions. This provides added incentives for accelerating the vaccine production cycle to reduce the lead time required for strain selection.

59 BASIC BIOLOGICAL SCIENCES↗

Quantifying Intramolecular Binding in Multivalent Interactions: A Structure-Based Synergistic Study on Grb2-Sos1 Complex

Numerous signaling proteins use multivalent binding to increase the specificity and affinity of their interactions within the cell. Enhancement arises because the effective binding constant for multivalent binding is larger than the binding constants for each individual interaction. We seek to gain both qualitative and quantitative understanding of the multivalent interactions of an adaptor protein, growth factor receptor bound protein-2 (Grb2), containing two SH3 domains interacting with the nucleotide exchange factor son-of-sevenless 1 (Sos1) containing multiple polyproline motifs separated by flexible unstructured regions. Grb2 mediates the recruitment of Sos1 from the cytosol to the plasma membrane where it activates Ras by inducing the exchange of GDP for GTP. First, using a combination of evolutionary information and binding energy calculations, we predict an additional polyproline motif in Sos1 that binds to the SH3 domains of Grb2. This gives rise to a total of five polyproline motifs in Sos1 that are capable of binding to the two SH3 domains of Grb2. Then, using a hybrid method combining molecular dynamics simulations and polymer models, we estimate the enhancement in local concentration of a polyproline motif on Sos1 near an unbound SH3 domain of Grb2 when its other SH3 domain is bound to a different polyproline motif on Sos1. We show that the local concentration of the Sos1 motifs that a Grb2 SH3 domain experiences is approximately 1000 times greater than the cellular concentration of Sos1. Finally, we calculate the intramolecular equilibrium constants for the crosslinking of Grb2 on Sos1 and use thermodynamic modeling to calculate the stoichiometry. With these equilibrium constants, we are able to predict the distribution of complexes that form at physiological concentrations. We believe this is the first systematic analysis that combines sequence, structure, and thermodynamic analyses to determine the stoichiometry of the complexes that are dominant in the cellular environment.

59 BASIC BIOLOGICAL SCIENCES↗

Making social networks more human: A topological approach

A key problem in social network analysis is to identify nonhuman interactions. State-of-the-art bot-detection systems like Botometer train machine-learning models on user-specific data. Unfortunately, these methods do not work on data sets in which only topological information is available. In this paper, we propose a new, purely topological approach. Our method removes edges that connect nodes exhibiting strong evidence of non-human activity from publicly available electronic-social-network datasets, including, for example, those in the Stanford Network Analysis Project repository (SNAP). Our methodology is inspired by classic work in evolutionary psychology by Dunbar that posits upper bounds on the total strength of the set of social connections in which a single human can be engaged. We model edge strength with Easley and Kleinberg's topological estimate; label nodes as “violators” if the sum of these edge strengths exceeds a Dunbar-inspired bound; and then remove the violator-to-violator edges. We run our algorithm on multiple social networks and show that our Dunbar-inspired bound appears to hold for social networks, but not for nonsocial networks. Our cleaning process classifies 0.04% of the nodes of the Twitter-2010 followers graph as violators, and we find that more than 80% of these violator nodes have Botometer scores of 0.5 or greater. Furthermore, after we remove the roughly 15 million violator-violator edges from the 1.2-billion-edge Twitter-2010 follower graph, 34% of the violator nodes experience a factor-of-two decrease in PageRank. PageRank is a key component of many graph algorithms such as node/edge ranking and graph sparsification. Thus, this artificial inflation would bias algorithmic output, and result in some incorrect decisions based on this output.

97 MATHEMATICS AND COMPUTING↗

DEVELOPMENT AND APPLICATION OF RISK ANALYSIS TOOLKIT FOR PLANT RESOURCE OPTIMIZATION

This paper presents the development of methods and tools that are being designed to optimize plant operations (e.g., maintenance/replacement schedules and optimal maintenance postures for plant components) in a manner that is more cost effective than current approaches and makes better use of available component health and cost data. These methods include both data- and model-based optimization methods. Model-based optimization methods directly include reliability and cost models to determine an optimal plant operational strategy. We consider gradient-based and evolutionary (based on genetic algorithms) optimization methods. The second class of methods target more specific use cases (e.g., project schedule optimization) and are not based on reliability models directly, but they require specific component reliability and cost data. This class of methods is based on variants of the knapsack problem with an aim to determine an optimal project schedule that maximizes the overall NPV. This paper also presents multi-objective methods designed to identify an optimal maintenance posture based on a Pareto frontier analysis. Rather than dictating the “right” tradeoff (i.e., identify the absolute best posture), we show how it is possible to perform a trade space exploration approach (i.e., identify value and costs of several postures and let the analysis account for desired value and cost metrics). This is performed by identifying maintenance postures that maximize value (e.g., system availability) and minimize operational costs, i.e., the Pareto frontier in a value-cost trade space. For all these methods we present detailed applicative examples that show their validity from a decision-making perspective.

97 - MATHEMATICS AND COMPUTING↗

A Method to Constrain Genome-Scale Models with 13 C Labeling Data

Current limitations in quantitatively predicting biological behavior hinder our efforts to engineer biological systems to produce biofuels and other desired chemicals. Here, we present a new method for calculating metabolic fluxes, key targets in metabolic engineering, that incorporates data from 13C labeling experiments and genome-scale models. The data from 13 C labeling experiments provide strong flux constraints that eliminate the need to assume an evolutionary optimization principle such as the growth rate optimization assumption used in Flux Balance Analysis (FBA). This effective constraining is achieved by making the simple but biologically relevant assumption that flux flows from core to peripheral metabolism and does not flow back. The new method is significantly more robust than FBA with respect to errors in genome-scale model reconstruction. Furthermore, it can provide a comprehensive picture of metabolite balancing and predictions for unmeasured extracellular fluxes as constrained by 13 C labeling data. A comparison shows that the results of this new method are similar to those found through 13 C Metabolic Flux Analysis ( 13 C MFA) for central carbon metabolism but, additionally, it provides flux estimates for peripheral metabolism. The extra validation gained by matching 48 relative labeling measurements is used to identify where and why several existing COnstraint Based Reconstruction and Analysis (COBRA) flux prediction algorithms fail. We demonstrate how to use this knowledge to refine these methods and improve their predictive capabilities. This method provides a reliable base upon which to improve the design of biological systems.

59 BASIC BIOLOGICAL SCIENCES↗

Online evolutionary neural architecture search for multivariate non-stationary time series forecasting

Time series forecasting (TSF) is one of the most important tasks in data science. TSF models are usually pre-trained with historical data and then applied on future unseen datapoints. However, real-world time series data is usually non-stationary and models trained offline usually face problems from data drift. Models trained and designed in an offline fashion can not quickly adapt to changes quickly or be deployed in real-time. To address these issues, this work presents the Online NeuroEvolution-based Neural Architecture Search (ONE-NAS) algorithm, which is a novel neural architecture search method capable of automatically designing and dynamically training recurrent neural networks (RNNs) for online forecasting tasks. Without any pre-training, ONE-NAS utilizes populations of RNNs that are continuously updated with new network structures and weights in response to new multivariate input data. ONE-NAS is tested on real-world, large-scale multivariate wind turbine data as well as the univariate Dow Jones Industrial Average (DJIA) dataset. These results demonstrate that ONE-NAS outperforms traditional statistical time series forecasting methods, including online linear regression, fixed long short-term memory (LSTM) and gated recurrent unit (GRU) models trained online, as well as state-of-the-art, online ARIMA strategies. Additionally, results show that utilizing multiple populations of RNNs which are periodically repopulated provide significant performance improvements, allowing this online neural network architecture design and training to be successful.

97 MATHEMATICS AND COMPUTING↗