Search NASA⌕ Search

SEARCH · Search NASA

Results for “machine-learned potentials”

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 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
NASA NTRS2021

Applying Machine Learning to Predict Alaskan Ionospheric Irregularities

In this work several machine-learning (ML) techniques for predicting ionospheric irregularities in the northern auroral zone were tested. The techniques include Ridge Regression, Long Short-Term Memory Neural Network (LSTM), Classification Neural Network (CNN), Autoencoder Classification Neural Network (ACNN), and LSTM Autoencoder Classification Neural Network (LACNN). These techniques were tested with the rate of total electron content (TEC) index (ROTI) data collected during 2008 and 2009 from a geodetic station in Fairbanks, Alaska (64.98°N, 147.50°W), which is in the auroral zone. Using ROTI data with the ML techniques, experiments were conducted to reach two goals: (1) examine what space weather measurements present good correlation with ROTI so that they may be helpful in ML-based prediction of ionospheric irregularities in the polar region; (2) predict ROTI hours and days ahead by training the neural network models with historical ROTI data alone. The Ridge Regression experiments indicate that a combination of measurements of local geomagnetic horizontal components, geomagnetic SYM-H index, 3-hour Kp and ap indices, and F10.7 solar flux index appears to be more correlated to the single-site ROTI measurements than other parameters. The neural network (NN) experiments show that although the LACNN model allows for predictions of non-irregularity and irregularity conditions defined by ROTI levels up to 3 hours in advance, with an overall accuracy ≥ 92%, a number of irregularity events can still be missed. Hence, further development is needed to reduce the number of missed events. In this paper, the models, data processing, model performance, prediction results, and potential applications are presented.

Pi, Xiaoqing↗
NASA NTRSDate not supplied

Do Better Satellite Precipitation Algorithms Improve Landslide Hazard Assessment?

Satellites make it possible to estimate precipitation in near real time. Given the challenges of achieving global coverage by other means, these data are used widely. However, few systems for landslide hazard assessment rely on satellite precipitation estimates. This could be due in part to perceptions of accuracy, although latency, spatial resolution, and other factors may also be important. We test whether recent changes to data streams from the Global Precipitation Measurement mission (GPM) have improved its potential for use in landslide prediction. Specifically, we examine data produced by the Integrated Multi-satellitERetrievals for the GPM (IMERG) algorithm, which was upgraded to version 7 this year. IMERG relies upon other algorithms, including the Goddard Profiling Algorithm (GPROF) and the GPM Combined Radar-Radiometer Algorithm (CORRA). Many changes have been made during the switch from IMERG version 6 to version 7. These include upgrading CORRA and GPROF to version 7, to improve the accuracy of precipitation in frozen, mountainous, and coastal areas. The measured intensity of some storms has been enhanced with a new algorithm, the Scheme for Histogram Adjustment with Ranked Precipitation Estimates in the Neighborhood. Combined with many others, these changes to IMERG should improve its utility for landslide hazard assessment in a variety of contexts. To test this idea, we retrain the global Landslide Hazard Assessment for Situational Awareness (LHASA) model twice—first with data from IMERG version 6B and second with 7B. Since current daily rainfall is the most important variable in determining outcomes predicted by LHASA, it should reflect changes made to that input. First, we grid the landslides at a daily, thirty-arcsecond resolution. This serves as the response variable. At each of these sites current and antecedent rainfall are extracted, along with antecedent snow mass and soil moisture, slope, and PGA. In addition, one million grid cells are selected at random points to represent conditions under which landslides (probably) do not occur. After merging these data, we hold back 20% of the dataset for validation purposes and train a machine-learning model with the rest. We assess both the model’s overall ability to identify landslides and its ability to predict specific large landslide disasters.

Thomas A Stanley↗
DOE OSTI2025

Even Higher-Level Synthesis: An Exploration of AI Hardware Accelerators using HLS4ML

With the rise of artificial intelligence, the popularization of deep learning, and a constantly evolving industry, the demand for flexible and efficient tools has never been greater. As algorithms grow more complex, their runtime and energy consumption increase exponentially. Customized hardware accelerators, long used for specific mathematical operations, remain essential for managing modern applications' computational and power demands. Hardware accelerators can speed up complex computations by orders of magnitude, but their manual design and verification processes are often challenging and time-consuming. High-Level Synthesis (HLS) provides a solution by transforming high-level algorithm descriptions, typically written in C++ or SystemC, into synthesizable RTL suitable for hardware implementation. This approach reduces development time for RTL engineers while offering flexibility beyond what traditional handwritten RTL can provide. We extended this capability to the machine-learning domain with the open-source framework hls4ml, which allows neural networks trained in Python frameworks like Tensorflow or PyTorch to be synthesized into efficient hardware representations for the traditional FPGA and ASIC flows. This breakthrough addresses the growing need for reduced design turnaround and easy verification of ML hardware accelerators with low latency and power efficiency constraints. During this tutorial, we will demonstrate how Python complements HLS by simplifying the ML design process, bridging the gap between software and hardware development. Attendees will explore how we translate neural networks modeled in Python into fixed-point C++ models suitable for HLS workflows. We will dive into strategies like Value-Range Analysis and Quantization-Aware Training, which optimize these designs for deployment and evaluate their accuracy, power consumption, and energy efficiency. To exemplify these concepts, experts from Fermilab will share their experiences applying this technology to high-energy physics experiments, where real-time, low-latency processing is critical. Over the years, Fermilab engineers have demonstrated how deep neural networks, optimized for hardware using hls4ml, can meet the stringent requirements of trigger systems at the CERN Large Hadron Collider. These systems rely on rapid decision-making to process immense data volumes while retaining only the most relevant events for further analysis. The application of hls4ml has also been extended to innovative technologies like smart pixel arrays. These smart pixels integrate ML inference capabilities directly into sensor devices, enabling localized data processing at the pixel level. This approach drastically reduces the need to transmit raw data to external processing units, significantly decreasing power consumption and latency. By embedding neural networks within the pixel architecture, the smart pixels can identify and prioritize relevant data in real time, providing a highly efficient solution for edge computing in scenarios such as particle detectors and imaging systems. Fermilab's work highlights the potential of hardware-accelerated ML in scenarios where both speed and power efficiency are mission-critical. Through this tutorial, attendees will gain valuable insights into the challenges and solutions of deploying ML in hardware. Understanding how HLS and hls4ml streamline the development of neural network-based hardware accelerators is fundamental for the industry's future. Participants will learn how these technologies are shaping the future of AI and scientific computing.

Di Guglielmo, Giuseppe [Fermilab]↗
NASA NTRS2020

A Machine Learning-Based Cloud Detection and Thermodynamic Phase Classification Algorithm using Passive Spectral Observations

We trained two Random Forest (RF) machine-learning models for cloud mask and cloud thermodynamic phase detection using spectral observations from VIIRS on Suomi NPP (SNPP). Observations from CALIOP were carefully selected to provide reference labels. The two RF models were trained for all-day and daytime-only conditions using a 4-year collocated VIIRS/CALIOP dataset from 2013 to 2016. Due to the orbit difference, the collocated CALIOP and SNPP VIIRS training samples cover a broad viewing zenith angle range, which is a great benefit to overall model performance. The all-day model uses 3 VIIRS infrared (IR) bands (8.6,11, and 12 μm) and the daytime model uses 5 Near-IR (NIR) and Shortwave-IR (SWIR) bands (0.86, 1.24, 1.38, 1.64 and 2.25 μm) together with the 3 IR bands to detect clear, liquid water, and ice cloud pixels. Up to 7 surface types, namely, ocean/water, forest, cropland, grassland, snow/ice, barren/desert, and shrubland, were considered separately to enhance performance for both models. Detection of cloudy pixels and thermodynamic phase with the two RF models were compared against collocated CALIOP products from 2017. It is shown that, with a conservative screening process that excludes the most challenging cloudy pixels for passive remote sensing, the two RF models have high accuracy rates in comparison with the CALIOP reference for both cloud detection and thermodynamic phase. Other existing SNPP VIIRS and Aqua MODIS cloud mask and phase products are also evaluated, with results showing that the two RF models and the MODIS MYD06 optical property phase product are the top 3 algorithms with respect to lidar observations during the daytime. During the nighttime, the RF all-day model works best for both cloud detection and phase, in particular for pixels over snow/ice surfaces. The present RF models can be extended to other similar passive instruments if training samples can be collected from CALIOP or other lidars. However, the quality of reference labels and potential sampling issues that may impact model performance would need further attention.

cloud detection↗
DOE OSTI2026

Crowdsourcing the Frontier: Advancing Hybrid Physics‐ML Climate Simulation via a $\$$50,000 Kaggle Competition

Subgrid machine-learning (machine learning [ML]) parameterizations have the potential to introduce a new generation of climate models that incorporate the effects of higher-resolution physics without incurring the prohibitive computational cost associated with more explicit physics-based simulations. However, important issues, ranging from online instability to inconsistent online performance, have limited their operational use for long-term climate projections. To more rapidly drive progress in solving these issues, domain scientists and ML researchers opened up the offline aspect of this problem to the broader ML and data science community with the release of ClimSim, a NeurIPS Data sets and Benchmarks publication, and an associated Kaggle competition. This paper reports on the downstream results of the Kaggle competition by coupling emulators inspired by the winning teams' architectures to an interactive climate model (including full cloud microphysics, a regime historically prone to online instability) and systematically evaluating their online performance. Our results demonstrate that online stability in the low-resolution real-geography setting is reproducible across multiple diverse architectures, which we consider a key milestone. All tested architectures exhibit strikingly similar offline and online biases, though their responses to architecture-agnostic design choices (e.g., expanding the list of input variables) can differ significantly. Multiple Kaggle-inspired architectures achieve state-of-the-art results on certain metrics such as zonal mean bias patterns and global Root Mean Squared Error, indicating that crowdsourcing the essence of the offline problem is one path to improving online performance in hybrid physics-AI climate simulation.

Environmental sciences↗
DOE OSTI2024

The ab initio non-crystalline structure database: empowering machine learning to decode diffusivity

Non-crystalline materials exhibit unique properties that make them suitable for various applications in science and technology, ranging from optical and electronic devices and solid-state batteries to protective coatings. However, data-driven exploration and design of non-crystalline materials is hampered by the absence of a comprehensive database covering a broad chemical space. In this work, we present the largest computed non-crystalline structure database to date, generated from systematic and accurate ab initio molecular dynamics (AIMD) calculations. We also show how the database can be used in simple machine-learning models to connect properties to composition and structure, here specifically targeting ionic conductivity. These models predict the Li-ion diffusivity with speed and accuracy, offering a cost-effective alternative to expensive density functional theory (DFT) calculations. Furthermore, the process of computational quenching non-crystalline structures provides a unique sampling of out-of-equilibrium structures, energies, and force landscape, and we anticipate that the corresponding trajectories will inform future work in universal machine learning potentials, impacting design beyond that of non-crystalline materials. In addition, combining diffusion trajectories from our dataset with models that predict liquidus viscosity and melting temperature could be utilized to develop models for predicting glass-forming ability.

36 MATERIALS SCIENCE↗
DOE OSTI2024

On ultrafast x-ray scattering methods for magnetism

With the introduction of x-ray free electron laser sources around the world, new scientific approaches for visualizing matter at fundamental length and time-scales have become possible. As it relates to magnetism and ‘magnetic-type’ systems, advanced methods are being developed for studying ultrafast magnetic responses on the time-scales at which they occur. We describe three capabilities which have the potential to seed new directions in this area and present original results from each: pump-probe x-ray scattering with low energy excitation, x-ray photon fluctuation spectroscopy, and ultrafast diffuse x-ray scattering. By combining these experimental techniques with advanced modeling together with machine learning, we describe how the combination of these domains allows for a new understanding in the field of magnetism. Finally, we give an outlook for future areas of investigation and the newly developed instruments which will take us there.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2024

Hamiltonian learning using machine-learning models trained with continuous measurements

Here, we build upon recent work on the use of machine-learning models to estimate Hamiltonian parameters using continuous weak measurement of qubits as input. We consider two settings for the training of our model: (1) supervised learning, where the weak-measurement training record can be labeled with known Hamiltonian parameters, and (2) unsupervised learning, where no labels are available. The first has the advantage of not requiring an explicit representation of the quantum state, thus potentially scaling very favorably to a larger number of qubits. The second requires the implementation of a physical model to map the Hamiltonian parameters to a measurement record, which we implement using an integrator of the physical model with a recurrent neural network to provide a model-free correction at every time step to account for small effects not captured by the physical model. We test our construction on a system of two qubits and demonstrate accurate prediction of multiple physical parameters in both the supervised context and the unsupervised context. We demonstrate that the model benefits from larger training sets, establishing that it is “learning,” and we show robustness regarding errors in the assumed physical model by achieving accurate parameter estimation in the presence of unanticipated single-particle relaxation.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Anomaly detection in collider physics via factorized observables

To maximize the discovery potential of high-energy colliders, experimental searches should be sensitive to unforeseen new physics scenarios. This goal has motivated the use of machine learning for unsupervised anomaly detection. In this paper, we introduce a new anomaly detection strategy called : factorized observables for regressing conditional expectations. Our approach is based on the inductive bias of factorization, which is the idea that the physics governing different energy scales can be treated as approximately independent. Assuming factorization holds separately for signal and background processes, the appearance of nontrivial correlations between low- and high-energy observables is a robust indicator of new physics. Under the most restrictive form of factorization, a machine-learned model trained to identify such correlations will in fact converge to the optimal new physics classifier. We test on a benchmark anomaly detection task for the Large Hadron Collider involving collimated sprays of particles called jets. By teasing out correlations between the kinematics and substructure of jets, our method can reliably extract percent-level signal fractions. This strategy for uncovering new physics adds to the growing toolbox of anomaly detection methods for collider physics with a complementary set of assumptions. Published by the American Physical Society 2024

Astronomy & Astrophysics↗
DOE OSTI2024

Next-Generation Materials Design: Quantum Mechanics and Data-Driven Modeling

The future of materials design is rapidly advancing through the combination of quantum mechanics and data-driven modeling. These approaches integrate quantum principles with advanced data analysis, enabling precise insights into material behavior. This talk will highlight recent progress in using these methods for computational design, particularly in high-entropy alloy catalysts, emphasizing the role of hierarchical machine-learning architectures for accurate predictions. Additionally, I will discuss our work on developing machine learning interatomic potentials (MLPs) for single-element metals, metal oxides, and alloys under extreme conditions, focusing on melting behavior and phase properties at high temperatures and pressures. We have also refined our MLP models to capture dynamic surface interactions, such as CO2 and CO adsorption on MgO, using both static and molecular dynamics simulations. These models maintain high accuracy while significantly reducing computational costs compared to first-principles calculations. By enabling efficient and accurate simulations, this work supports broader community adoption, optimizes datasets for materials discovery, and extends the accessible time, size, and environmental conditions beyond the limits of experiments and traditional simulations.

machine learning↗
DOE OSTI2026

Investigation of the Effect of Framework Flexibility on Adsorption in SIFSIX-3-Cu using a Machine-Learned Force Field

Metal-organic frameworks (MOFs) are a promising class of adsorbents. The performance of MOF sorbents relies on high selectivity and low regeneration energy. This work focuses on the use of machine learned force fields (MLFFs) to model adsorption in a flexible MOF, SIFSIX-3-Cu. A DeePMD-based MLFF was trained to reproduce DFT (PBE+D3) energies, forces, and stresses, using an iterative sampling scheme combining sampling based on molecular dynamics, Monte Carlo, and geometry optimization to capture both attractive and repulsive regions of the potential energy surface. Flexibility of the MOF was explicitly included in this model. Hybrid Monte Carlo/molecular dynamics (MC/MD) simulations using the MLFF predicted adsorption isotherms in good agreement with experimental data for a range of pressures (40 Pa – 104 Pa) in contrast to rigid models, which overpredict CO2 adsorption at low pressures. The improvement was the result of a description of the variability of fluorine-fluorine diagonal distances at adsorption sites. This detailed description of flexibility afforded by the MLFF resulted in more accurate predictions adsorption isotherms when compared to the experimentally measured values. These results underscore the importance of including framework flexibility when modeling adsorption phenomena in MOFs, particularly for low pressure applications and provide a robust procedure for training MLFF models for MOFs.

Atomistic Simulation↗
NASA NTRS2013

Modeling Weather Impact on Airport Arrival Miles-in-Trail Restrictions

When the demand for either a region of airspace or an airport approaches or exceeds the available capacity, miles-in-trail (MIT) restrictions are the most frequently issued traffic management initiatives (TMIs) that are used to mitigate these imbalances. Miles-intrail operations require aircraft in a traffic stream to meet a specific inter-aircraft separation in exchange for maintaining a safe and orderly flow within the stream. This stream of aircraft can be departing an airport, over a common fix, through a sector, on a specific route or arriving at an airport. This study begins by providing a high-level overview of the distribution and causes of arrival MIT restrictions for the top ten airports in the United States. This is followed by an in-depth analysis of the frequency, duration and cause of MIT restrictions impacting the Hartsfield-Jackson Atlanta International Airport (ATL) from 2009 through 2011. Then, machine-learning methods for predicting (1) situations in which MIT restrictions for ATL arrivals are implemented under low demand scenarios, and (2) days in which a large number of MIT restrictions are required to properly manage and control ATL arrivals are presented. More specifically, these predictions were accomplished by using an ensemble of decision trees with Bootstrap aggregation (BDT) and supervised machine learning was used to train the BDT binary classification models. The models were subsequently validated using data cross validation methods. When predicting the occurrence of arrival MIT restrictions under low demand situations, the model was able to achieve over all accuracy rates ranging from 84% to 90%, with false alarm ratios ranging from 10% to 15%. In the second set of studies designed to predict days on which a high number of MIT restrictions were required, overall accuracy rates of 80% were achieved with false alarm ratios of 20%. Overall, the predictions proposed by the model give better MIT usage information than what has been currently provided under current day operations. Traffic flow managers can use these predictions to identify potential MIT restrictions to eliminate (e.g., those occurring during low arrival demand periods), and to determine the days in which a significant number of restrictions may be required

Operation↗
NASA NTRS2016

Multi-Objective Reinforcement Learning for Cognitive Radio-Based Satellite Communications

Previous research on cognitive radios has addressed the performance of various machine-learning and optimization techniques for decision making of terrestrial link properties. In this paper, we present our recent investigations with respect to reinforcement learning that potentially can be employed by future cognitive radios installed onboard satellite communications systems specifically tasked with radio resource management. This work analyzes the performance of learning, reasoning, and decision making while considering multiple objectives for time-varying communications channels, as well as different cross-layer requirements. Based on the urgent demand for increased bandwidth, which is being addressed by the next generation of high-throughput satellites, the performance of cognitive radio is assessed considering links between a geostationary satellite and a fixed ground station operating at Ka-band (26 GHz). Simulation results show multiple objective performance improvements of more than 3.5 times for clear sky conditions and 6.8 times for rain conditions.

software defined radio↗
NASA NTRSDate not supplied

Toward Design Assurance of Machine-Learning Airborne Systems

In recent years, Artificial Intelligence (AI) systems, enabled by Machine Learning (ML)technology, have demonstrated impressive progress and provides historic opportunities for the aviation industry. However, several key aspects of ML technology are not compatible with existing design assurance standards and make certification problematic. In this paper, we present a case study of a visual system with a Deep Neural Network (DNN) intended to detect and identify airport runway signs. Different use cases and variants of this system exhibit different levels of criticality ranging from design assurance level (DAL) D to B. We use the case study to illustrate the challenges of certification according to the current standards, such asDO-178C. We present the system design, data generation, training, and verification in detail and describe how the design assurance objectives can be met for a DAL D variant of the system. We also discuss gaps and potential approaches for the higher design assurance levels.

Avionics↗
NASA NTRS2021

Machine Learning Models to Predict Cognitive Impairment of Rodents Subjected to Space Radiation

This research uses machine-learned computational analyses to predict the cognitive performance impairment of rats induced by irradiation. The experimental data in the analyses is from a rodent model exposed to ≤ 15 cGy of individual Galactic Cosmic Radiation (GCR) ions: 4He, 16O, 28Si, 48Ti, or 56Fe, expected for a Lunar or Mars mission. This work investigates rats at a subject-based level and uses performance scores taken before irradiation to predict impairment in Attentional Set-shifting (ATSET) data post-irradiation. Here, the worst performing rats of the control group define the impairment thresholds based on population analyses via cumulative distribution functions, leading to the labeling of impairment for each subject. A significant finding is the exhibition of a dose-dependent increasing probability of impairment for 1 to 10 cGy of 28Si or 56Fe in the Simple Discrimination (SD) stage of the ATSET, and for 1 to 10 cGy of 56Fe in the Compound Discrimination (CD) stage. On a subject-based level, implementing Machine Learning (ML) classifiers such as the Gaussian Naïve Bayes, Support Vector Machine, and Artificial Neural Networks identifies rats that have a higher tendency for impairment after GCR exposure. The algorithms employ the experimental prescreenperformance scores as multidimensional input features to predict each rodent’s susceptibility to cognitive impairment due to space radiation exposure. The receiver operating characteristic and the precision-recall curves of the ML models show a better prediction of impairment when 56Feis the ion in question in both SD and CD stages. They, however, do not depict impairment due to 4Hein SD and 28Siin CD, suggesting no dose-dependent impairment response in these cases. One key finding of our study is that prescreen performance scores can be used to predict the ATSET performance impairments. This result is significant to crewed space missions as it supports the potential of predicting an astronaut’s impairment in a specific task before spaceflight through the implementation of appropriately trained ML tools. Future research can focus on constructing ML ensemble methods to integrate the findings from the methodologies implemented in this study for morerobust predictionsof cognitive decrements due to space radiation exposure.

space radiation↗
NASA NTRS2022

Classifying Unidentified X-Ray Sources in the Chandra Source Catalog Using A Multiwavelength Machine-Learning Approach

The rapid increase in serendipitous X-ray source detections requires the development of novel approaches to efficiently explore the nature of X-ray sources. If even a fraction of these sources could be reliably classified, it would enable population studies for various astrophysical source types on a much larger scale than currently possible. Classification of large numbers of sources from multiple classes characterized by multiple properties (features) must be done automatically and supervised machine learning (ML) seems to provide the only feasible approach. We perform classification of Chandra Source Catalog version 2.0 (CSCv2) sources to explore the potential of the ML approach and identify various biases, limitations, and bottlenecks that present themselves in these kinds of studies. We establish the framework and present a flexible and expandable Python pipeline, which can be used and improved by others. We also release the training data set of 2941 X-ray sources with confidently established classes. In addition to providing probabilistic classifications of 66,369 CSCv2 sources (21% of the entire CSCv2 catalog), we perform several narrower-focused case studies (high-mass X-ray binary candidates and X-ray sources within the extent of the H.E.S.S. TeV sources) to demonstrate some possible applications of our ML approach. We also discuss future possible modifications of the presented pipeline, which are expected to lead to substantial improvements in classification confidences.

Hui Yang↗
NASA NTRSDate not supplied

Evaluating Meteorological Dust Events and Machine-Learning Based Dust Identification in Geostationary Satellite Imagery

NASA scientists in the Short-term Prediction Research and Transition Center (SPoRT) developed a physically-based machine learning approach to identify dust in satellite imagery with a focus on night-time dust detection (Berndt et al. 201; DustTracker-AI). NASA/NOAA Geostationary Environmental Operational Satellite-16 (GOES-16) imagery was used for training and model inputs. The training, testing and validation data set consists of 28 events in the Southwest United States, capturing dust and null events in the region from 2018-2020.With 83 distinct images and millions of pixels a random forest model was trained and validated, correctly labeling 85% of dust pixels.For the first time, the model was run in near-real time production during the spring of 2022 and dust probability visualizations were made available to NOAA National Weather Service (NWS) forecasters to assess its utility for dust forecasting. Results indicated the model helped increase the confidence in the presence of dust and enabled dust tracking for a longer period of time into the night-time hours. Forecaster assessment and running the model in near real-time allowed for the team to determine the types of events missed, captured, and false alarms. To gain additional context on model performance,the SPoRT team sought to gather more detailed information on the training database(e.g., meteorological characteristics and drivers). The goal of this project was to identify the meteorological drivers for the dust events and create a database which synthesized information from observations, forecaster discussions, and analyses pertaining to the dust events to understand the types of events currently used to train the model. A more detailed meteorological synopsis was created for each dust event in the training, testing, and validation datasets. Following the completion of the database and documentation, the classification details revealed that 88% of the dust events were synoptically driven while mesoscale events were less prevalent in model datasets. Meteorological conditions found such as mixing layer depth and wind velocity had mean values of 645mb and 21kt respectively.With conditions of deep mixed layers and moderate to strong surface winds a mesoscale thunderstorm outflow event was considered and subsequently added to the model training data set to test the impact of additional mesoscale training data. The model was retrained and then qualitatively tested on a sample thunderstorm outflow case that the original model was unable to identify. Preliminary results showed potential that the addition of more mesoscale events included in the training data could help to better identify indistinct and localized dust events.

Connor Welch↗
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-of-the-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.

Deep-learning↗