Search NASA⌕ Search

SEARCH · Search NASA

Results for “Feature selection”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 55 records · Page 3
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Preprocessing for Unintended Conducted Emissions Classification with ResNet

Characterization of Unintended Conducted Emissions (UCE) from electronic devices is important when diagnosing electromagnetic interference, performing nonintrusive load monitoring (NILM) of power systems, and monitoring electronic device health, among other applications. Prior work has demonstrated that UCE analysis can serve as a diagnostic tool for energy efficiency investigations and detailed load analysis. While explaining the feature selection of deep networks with certainty is often not fully comprehensive, or in other applications, quite lacking, additional tools/methods for further corroboration and confirmation can help further the understanding of the researcher. This is true especially in the subject application of the study in this paper. Often the focus of such efforts is the selected features themselves, and there is not as much understanding gained about the noise in the collected data. If selected feature and noise characteristics are known, it can be used to further shape the design of the deep network or associated preprocessing. This is additionally difficult when the available data are limited, as in the case which the authors investigated in this study. Here, the authors present a novel work (which is a proposed complementary portion of the overall solution to the deep network classification explainability problem for this application) by applying a systematic progression of preprocessing and a deep neural network (ResNet architecture) to classify UCE data obtained via current transformers. By using a methodical application of preprocessing techniques prior to a deep classifier, hypotheses can be produced concerning what features the deep network deems important relative to what it perceives as noise. For instance, it is hypothesized in this particular study as a result of execution of the proposed method and periodic inspection of the classifier output that the UCE spectral features are relatively close to each other or to the interferers, as systematically reducing the beta parameter of the Kaiser window produced progressively better classification performance, but only to a point, as going below the Beta of eight produced decreased classifier performance, as well as the hypothesis that further spectral feature resolution was not as important to the classifier as rejection of the leakage from a spectrally distant interference. This can be very important in unpredictable low-FNR applications, where knowing the difference between features and noise is difficult. As a side-benefit, much was learned regarding the best preprocessing to use with the selected deep network for the UCE collected from these low power consumer devices obtained via current transformers. Baseline rectangular windowed FFT preprocessing provided a 62% classification increase versus using raw samples. After performing a more optimal preprocessing, more than 90% classification accuracy was achieved across 18 low-power consumer devices for scenarios in which the in-band features-to-noise ratio (FNR) was very poor.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2024

Geographical Insights into Suicide Mortality Through Spatial Machine Learning

Suicide mortality is a leading cause of death in the United States, with an upward trend that emphasizes its significance as a public health issue. Previous research has employed global models like ordinary least squares (OLS) regression and local models such as geographically weighted regression (GWR). While local models are useful for analyzing spatial variations in suicide mortality, they share limitations with traditional global models, particularly about their inability to handle multi-collinearity and non-linear relationships. Machine learning approaches, like random forests (RF), can address some of these limitations but often fail to account for spatial variability. This gap highlights the need for spatial ML models specifically designed to tackle suicide mortality. This research seeks to fill this void by using a geographically weighted random forest model (GWRF) to examine the associations between county-level suicide mortality in the U.S. from 2010 to 2020 and various social and environmental determinants of health. A key aspect of our methodology is disciplined feature selection, which reduces the pool of explanatory variables by about 90%. This refinement enhances the explanatory power of both global (R2 improved from 0.59 to 0.67) and local (R2 improved from 0.64 to 0.67) RF models while reducing their run times. An analysis of the importance scores for these selected features reveals that the drivers of suicide mortality vary by context. Thus, to effectively address regional disparities and inform targeted public health interventions, a holistic approach that incorporates multiple county-level characteristics is essential.

Lebakula, Viswadeep [ORNL] (ORCID:0000000152935914↗
DOE OSTI2022

A systematic method for selecting molecular descriptors as features when training models for predicting physiochemical properties

Machine learning has proven to be a powerful tool for accelerating biofuel development. Although numerous models are available to predict a range of properties using chemical descriptors, there is a trade-off between interpretability and performance. Neural networks provide predictive models with high accuracy at the expense of some interpretability, while simpler models such as linear regression often lack in accuracy. In addition to model architecture, feature selection is also critical for developing interpretable and accurate predictive models. We present a method for systematically selecting molecular descriptor features and developing interpretable machine learning models without sacrificing accuracy. Our method simplifies the process of selecting features by reducing feature multicollinearity and enables discoveries of new relationships between global properties and molecular descriptors. To demonstrate our approach, we developed models for predicting melting point, boiling point, flash point, yield sooting index, and net heat of combustion with the help of the Tree-based Pipeline Optimization Tool (TPOT). For training, we used publicly available experimental data for up to 8351 molecules. Our models accurately predict various molecular properties for organic molecules (mean absolute percent error (MAPE) ranges from 3.3% to 10.5%) and provide a set of features that are well-correlated to the property. This method enables researchers to explore sets of features that significantly contribute to the prediction of the property, offering new scientific insights. To help accelerate early stage biofuel research and development, we also integrated the data and models into a open-source, interactive web tool.

09 BIOMASS FUELS↗
DOE OSTI2026

Virtual refrigerant charge sensing algorithm for residential CO₂ heat pumps

Natural refrigerants are increasingly adopted in next-generation heat pump systems, among which CO₂ heat pumps have attracted significant attention. However, due to their high operating pressures, the leakage risk is higher, resulting in undercharge conditions and degraded heat pump performance. Thus, developing an accurate refrigerant charge level detection technique is necessary to guarantee safe and efficient operation. Although virtual refrigerant charge (VRC) level calculation algorithms for CO₂ heat pumps exist, they typically rely on empirically selected features without a systematic selection framework, leading to multicollinearity and potential overfitting, which limit their prediction accuracy and generalizability. To address these issues, this study proposes a VRC algorithm framework with a systematic feature selection method that identifies physically meaningful and statistically significant features, and is applied using a residential CO₂ heat pump as a case study. The method is extended from previous work on conventional refrigerants to account for charge behavior in CO₂ gas coolers. The selected features include gas cooler outlet density, evaporator pressure, and superheat temperature. The results demonstrate that the proposed feature selection method significantly improves prediction accuracy compared to existing VRC approaches. A relatively small training dataset (∼30 samples) is sufficient for feature identification and model development. The developed algorithm achieves less than 3% prediction error under both undercharge and overcharge conditions, representing reductions of 46.7% and 35.3% compared to two recent reference VRC algorithms for transcritical CO₂ heat pumps reported in the literature. The proposed algorithm and feature selection method enhance leakage detection capability, facilitate the deployment of CO₂ heat pump systems, and contribute to reduced energy waste and maintenance costs.

Guo, Fangzhou [Lawrence Berkeley National Laborato↗
DOE OSTI2020

Clustering Analysis of Commercial Vehicles Using Automatically Extracted Features from Time Series Data

Standard of practice approaches to time series cluster analysis involve careful feature engineering, often utilizing expert input to tune and select features by hand. In many cases, expert input may not be readily available, or there may not yet exist a community consensus on the ideal features for a given application. This paper compares the results of several cluster analysis methods, using both hand selected features and those extracted automatically, when applied to large geospatial time series telematics data from commercial trucking fleets. The impacts of feature selection, dimensionality reduction, and choice of clustering algorithm on the quality of clustering results are explored. Results from this analysis confirm prior results that domain agnostic features are competitive with the hand engineered features with respect to clustering quality metrics. These results also provide new insight into the most successful strategies for identifying structure in large unstructured vehicle telematics data, and suggest that time series clustering using automatic feature extraction can be an effective approach to extract structure from large scale geospatial time series data in cases when hand selected features are not available.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2022

Analytics-at-scale of Sensor Data for Digital Monitoring in Nuclear Plants (3 rd Annual Report)

Nuclear power plants collect and store large volumes of heterogeneous data from various components and systems. With recent advances in machine learning (ML) techniques, these data can be leveraged to develop diagnostic and short-term forecasting models to better predict future equipment condition. Maintenance operations can then be planned in advance whenever degraded performance is predicted, thus resulting in fewer unplanned outages and the optimization of maintenance activities. This enables lower maintenance costs and improves the overall economics of nuclear power. This report primarily focuses on developing a short-term forecasting process that leverages a feature selection process to distill large volumes of heterogeneous data and predict specific equipment parameters. A variety of feature selection methods, including Shapley Additive Explanations (SHAP) and variance inflation factor (VIF), were used to select the optimal features as inputs for three ML methods: long short-term memory (LSTM) networks, support vector regression (SVR), and random forest (RF). Each combination of model and input features was used to predict a pump bearing temperature both 1 and 24 hours in advance, based on actual plant system data. The optimal inputs for the LSTM and SVR were selected using the SHAP values, while the optimal input for the RF consisted solely of the response variable itself. Each model produced similar 1-hour-ahead predictions, with root mean square errors (RMSEs) of roughly 0.006. For the 24-hour-ahead predictions, differences could be seen between LSTM, SVR, and RF, as reflected by model performances of 0.036 ± 0.014, 0.0026 ± 0, and 0.063 ± 0.004 RMSE, respectively. As big data and continuous online monitoring become more widely available, the proposed feature selection process can be used for many applications beyond the prediction of process parameters within nuclear infrastructure. This report summarizes the Fiscal Year 2021 research progress encompassing the (1) data cleaning and feature selection necessary for ML applications; (2) development of short-term forecasting models to predict future plant process parameters for both single and multiple time steps ahead; and (3) validation of the feature selection methods and short-term forecasting models given new data from different systems.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2020

Quantifying Leaf Chlorophyll Concentration of Sorghum from Hyperspectral Data Using Derivative Calculus and Machine Learning

Leaf chlorophyll concentration (LCC) is an important indicator of plant health, vigor, physiological status, productivity, and nutrient deficiencies. Hyperspectral spectroscopy at leaf level has been widely used to estimate LCC accurately and non-destructively. This study utilized leaf-level hyperspectral data with derivative calculus and machine learning to estimate LCC of sorghum. We calculated fractional derivative (FD) orders starting from 0.2 to 2.0 with 0.2 order increments. Additionally, 43 common vegetation indices (VIs) were calculated from leaf spectral reflectance factor to make comparisons with reflectance-based data. Within the modeling pipeline, three feature selection methods were assessed: Pearson’s correlation coefficient (PCC), partial least squares based variable importance in the projection (VIP), and random forest-based mean decrease impurity (MDI). Finally, we used partial least squares regression (PLSR), random forest regression (RFR), support vector regression (SVR), and extreme learning regression (ELR) to estimate the LCC of sorghum. Results showed that: (1) increasing derivative order can show improved model performance until certain order for reflectance-based analysis; however, it is inconclusive to state that a particular order is optimal for estimating LCC of sorghum; (2) VI-based modeling outperformed derivative augmented reflectance factor-based modeling; (3) mean decrease impurity was found effective in selecting sensitive features from large feature space (reflectance-based analysis), whereas simple Pearson’s correlation coefficient worked better with smaller feature space (VI-based analysis); and (4) SVR outperformed all other models within reflectance-based analysis; alternatively, ELR with VIs from original reflectance yielded slightly better results compared to all other models.

47 OTHER INSTRUMENTATION↗
DOE OSTI2021

Clinical knowledge extraction via sparse embedding regression (KESER) with multi-center large scale electronic health record data

The increasing availability of electronic health record (EHR) systems has created enormous potential for translational research. However, it is difficult to know all the relevant codes related to a phenotype due to the large number of codes available. Traditional data mining approaches often require the use of patient-level data, which hinders the ability to share data across institutions. In this project, we demonstrate that multi-center large-scale code embeddings can be used to efficiently identify relevant features related to a disease of interest. We constructed large-scale code embeddings for a wide range of codified concepts from EHRs from two large medical centers. We developed knowledge extraction via sparse embedding regression (KESER) for feature selection and integrative network analysis. We evaluated the quality of the code embeddings and assessed the performance of KESER in feature selection for eight diseases. Besides, we developed an integrated clinical knowledge map combining embedding data from both institutions. The features selected by KESER were comprehensive compared to lists of codified data generated by domain experts. Features identified via KESER resulted in comparable performance to those built upon features selected manually or with patient-level data. The knowledge map created using an integrative analysis identified disease-disease and disease-drug pairs more accurately compared to those identified using single institution data. Analysis of code embeddings via KESER can effectively reveal clinical knowledge and infer relatedness among codified concepts. KESER bypasses the need for patient-level data in individual analyses providing a significant advance in enabling multi-center studies using EHR data.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗
DOE OSTI2022

Numerical characterization of support recovery in sparse regression with correlated design

Sparse regression is employed in diverse scientific settings as a feature selection method. A pervasive aspect of scientific data is the presence of correlations between predictive features. These correlations hamper both feature selection and estimation and jeopardize conclusions drawn from estimated models. On the other hand, theoretical results on sparsity-inducing regularized regression have largely addressed conditions for selection consistency via asymptotics, and disregard the problem of model selection, whereby regularization parameters are chosen. In this numerical study, we address these issues through exhaustive characterization of the performance of several regression estimators, coupled with a range of model selection strategies. These estimators and selection criteria were examined across correlated regression problems with varying degrees of signal to noise, distributions of non-zero model coefficients, and model sparsity. Our results reveal a fundamental tradeoff between false positive and false negative control in all regression estimators and model selection criteria examined. Additionally, we numerically explore a transition point modulated by the signal-to-noise ratio and spectral properties of the design covariance matrix at which the selection accuracy of all considered algorithms degrades. Overall, we find that SCAD coupled with BIC or empirical Bayes model selection performs the best feature selection across the regression problems considered.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

A data-driven framework for permeability prediction of natural porous rocks via microstructural characterization and pore-scale simulation

Understanding the microstructure–property relationships of porous media is of great practical significance, based on which macroscopic physical properties can be directly derived from measurable microstructural informatics. However, establishing reliable microstructure–property mappings in an explicit manner is difficult, due to the intricacy, stochasticity, and heterogeneity of porous microstructures. In this paper, a data-driven computational framework is presented to investigate the inherent microstructure–permeability linkage for natural porous rocks, where multiple techniques are integrated together, including microscopy imaging, stochastic reconstruction, microstructural characterization, pore-scale simulation, feature selection, and data-driven modeling. A large number of 3D digital rocks with a wide porosity range are acquired from microscopy imaging and stochastic reconstruction techniques. A broad variety of morphological descriptors are used to quantitatively characterize pore microstructures from different perspectives, and they compose the raw feature pool for feature selection. Here high-fidelity lattice Boltzmann simulations are conducted to resolve fluid flow passing through porous media, from which reliable permeability references are obtained. The optimal feature set that best represents permeability is identified through a performance-oriented feature selection process, upon which a cost-effective surrogate model is rapidly fitted to approximate the microstructure-permeability mapping via data-driven modeling. This surrogate model exhibits great advantages over empirical/analytical formulas in terms of prediction accuracy and generalization capacity, which can predict reliable permeability values spanning four orders of magnitude. Besides, feature selection also greatly enhances the interpretability of the data-driven prediction model, from which new insights into the mechanism of how microstructural characteristics determine intrinsic permeability are obtained.

58 GEOSCIENCES↗
DOE OSTI2025

JGI-Trichoderma v1.0

There is a series of Python and bash scripts to parse genomics datasets used to evaluate the coevolution of gene families and the feature importance of gene families using an SVM classifier. - Cover analysis: takes a list of single-copy genes in a set of genomes, aligns and builds the gene trees to determine if two gene families have a signature of covariation with one another. It parses the files to run phykit cover script described here: https://jlsteenwyk.com/PhyKIT/usage/index.html - SVM-classifier: This Python script is an SVM-based genomic classifier designed for biological data analysis. It combines machine learning with feature selection to identify important genomic markers and classify biological samples. Core Functionality: The script uses Support Vector Machines from scikit-learn to classify genomic data, incorporating SelectKBest for automated feature selection and leave-one-out cross-validation for performance assessment. It operates in multiple modes: feature ranking, optimal combination discovery, and sample prediction. Primary Applications: Genomic sample classification and biomarker discovery Feature importance analysis in high-dimensional biological datasets Prediction of sample categories based on genomic profiles Research applications requiring robust classification of biological data Key Advantages: High-dimensional handling: SVMs excel with genomic data's typical high feature-to-sample ratios Integrated feature selection: Reduces noise and computational overhead while identifying key markers Probability estimation: Provides confidence scores essential for biological interpretation Validation robustness: Leave-one-out cross-validation ensures reliable performance metrics Operational flexibility: Multiple analysis modes support different research phases from exploration to prediction

Stecca Steindorff, Andrei [Lawrence Berkeley Natio↗
DOE OSTI2023

Multilevel Robustness for 2D Vector Field Feature Tracking, Selection and Comparison

Abstract Critical point tracking is a core topic in scientific visualization for understanding the dynamic behaviour of time‐varying vector field data. The topological notion of robustness has been introduced recently to quantify the structural stability of critical points, that is, the robustness of a critical point is the minimum amount of perturbation to the vector field necessary to cancel it. A theoretical basis has been established previously that relates critical point tracking with the notion of robustness, in particular, critical points could be tracked based on their closeness in stability, measured by robustness, instead of just distance proximity within the domain. However, in practice, the computation of classic robustness may produce artifacts when a critical point is close to the boundary of the domain; thus, we do not have a complete picture of the vector field behaviour within its local neighbourhood. To alleviate these issues, we introduce a multilevel robustness framework for the study of 2D time‐varying vector fields. We compute the robustness of critical points across varying neighbourhoods to capture the multiscale nature of the data and to mitigate the boundary effect suffered by the classic robustness computation. We demonstrate via experiments that such a new notion of robustness can be combined seamlessly with existing feature tracking algorithms to improve the visual interpretability of vector fields in terms of feature tracking, selection and comparison for large‐scale scientific simulations. We observe, for the first time, that the minimum multilevel robustness is highly correlated with physical quantities used by domain scientists in studying a real‐world tropical cyclone dataset. Such an observation helps to increase the physical interpretability of robustness.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Is Knowledge About Running Applications Helping Improve Runtime Prediction of HPC Jobs?

High-performance computing systems rely upon scheduling algorithms to achieve high utilization. These schedulers rely upon user estimates of job resource requirements, such as runtime, to determine optimal scheduling of incoming jobs. These user estimates, however, are prone to error. To mitigate this error, significant research has been directed at providing better estimates of job runtime, usually employing machine learning techniques. These techniques are dependent upon the input features selected. Among the possible features is the primary application used by the job. In a survey of more than 20 papers directed at improving runtime prediction, only four included primary application as an input feature. We focus this investigation specifically on the value of adding primary application as an input feature, and find that it does improve model performance, especially for jobs with longer runtimes, though this improvement varies based on the application used. We recommend further research to determine the cause of this variability as well as an optimal strategy for employing a mixture of models both including and not including primary application as a feature.

feature selection↗
DOE OSTI2019

Benchmarking Parametric and Machine Learning Models for Genomic Prediction of Complex Traits

The usefulness of genomic prediction in crop and livestock breeding programs has prompted efforts to develop new and improved genomic prediction algorithms, such as artificial neural networks and gradient tree boosting. However, the performance of these algorithms has not been compared in a systematic manner using a wide range of datasets and models. Using data of 18 traits across six plant species with different marker densities and training population sizes, we compared the performance of six linear and six non-linear algorithms. First, we found that hyperparameter selection was necessary for all non-linear algorithms and that feature selection prior to model training was critical for artificial neural networks when the markers greatly outnumbered the number of training lines. Across all species and trait combinations, no one algorithm performed best, however predictions based on a combination of results from multiple algorithms (i.e., ensemble predictions) performed consistently well. While linear and non-linear algorithms performed best for a similar number of traits, the performance of non-linear algorithms vary more between traits. Although artificial neural networks did not perform best for any trait, we identified strategies (i.e., feature selection, seeded starting weights) that boosted their performance to near the level of other algorithms. Our results highlight the importance of algorithm selection for the prediction of trait values.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2019

Benchmarking parametric and machine learning models for genomic prediction of complex traits

The usefulness of genomic prediction in crop and livestock breeding programs has prompted efforts to develop new and improved genomic prediction algorithms, such as artificial neural networks and gradient tree boosting. However, the performance of these algorithms has not been compared in a systematic manner using a wide range of datasets and models. Using data of 18 traits across six plant species with different marker densities and training population sizes, we compared the performance of six linear and six non-linear algorithms. First, we found that hyperparameter selection was necessary for all non-linear algorithms and that feature selection prior to model training was critical for artificial neural networks when the markers greatly outnumbered the number of training lines. Across all species and trait combinations, no one algorithm performed best, however predictions based on a combination of results from multiple algorithms (i.e. ensemble predictions) performed consistently well. While linear and non-linear algorithms performed best for a similar number of traits, the performance of non-linear algorithms vary more between traits. Although artificial neural networks did not perform best for any trait, we identified strategies (i.e. feature selection, seeded starting weights) that boosted their performance to near the level of other algorithms. Our results highlight the importance of algorithm selection for the prediction of trait values.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2020

Time Series Classification for Locating Forced Oscillation Sources

Here, this article presents a machine learning based time-series classification method for using synchrophasor measurements to locate the source of forced oscillation (FO) for fast disturbance removal. First, multivariate time series (MTS) matrices are constructed by the most informative measurements selected by sequential feature selection from each power plant. Then, the Mahalanobis matrix is trained such that the Mahalanobis distance between the MTSs from the same class (i.e., with the same FO source location) are minimized and from different classes (i.e., with different FO source locations) are maximized. This allows MTSs to be classified by classifiers with class membership corresponding to the location of each FO source. To meet the runtime requirements of online matching, class templates are constructed to reduce data size and improve matching efficiency. To account for uncertainty in identifying the exact beginning of an FO event, dynamic time warping is used to align the out-of-sync MTSs. IEEE 39bus and WECC 179bus systems are used for algorithm development and validation. Simulation results demonstrate that the algorithm meets online operation runtime requirement with high accuracy using misaligned data sets.

42 ENGINEERING↗
DOE OSTI2020

A systematic feature extraction and selection framework for data-driven whole-building automated fault detection and diagnostics in commercial buildings

In data-driven automated fault detection and diagnostics (AFDD) modeling for building energy systems, feature engineering is a critical process of extracting information from high-dimensional and noisy sensor measurement and turning it into informative and representative inputs or features for data-driven modeling. However, few studies specifically discuss the feature engineering, especially the interactions between feature extraction and feature selection in whole-building AFDD. We developed a systematic feature extraction and selection framework for whole-building AFDD. In this framework, features are aggressively extracted from raw sensor data using statistical feature extraction techniques with various window sizes and statistics. With many features extracted, a hybrid feature selection algorithm that combines the filter and wrapper method then selects the best feature set. The framework considers diversity in the duration of fault behavior among fault types in whole-building AFDD, thus achieving high model generalization. We implemented our developed framework in a virtual testbed calibrated with measured data from Oak Ridge National Laboratory's Flexible Research Platform designed to mimic the operation of a typical small commercial building. The AFDD model is trained by the simulation data generated from the virtual testbed. The results show that (1) the developed framework improves the generalization of the AFDD model by 10.7% compared with literature-reported feature extraction and selection methods and (2) features with diverse window sizes and statistics are selected, providing insight into physical systems beyond the current understanding of buildings and faults and improving the detection and diagnostics of multiple fault types.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2019

Infrared and radar fusion detection method based on heterogeneous data preprocessing

The difficulty in infrared and radar fusion target detection lies in the pre-processing steps such as sample reduction and feature reduction of heterogeneous sensors. For the problem of data volume redundancy, HVS sample reduction is proposed to reduce the amount of data in the infrared data set. Infrared and radar joint feature vectors are constructed through nearest neighbor data association. For the problem of feature redundancy, a weighted mutual information feature selection method based on prior information is proposed according to the difference of feature sources. The fusion data set is then used for classification. The experiments show that the HVS-based sample reduction and the prior weighted feature selection achieved a higher detection probability and a shorter detection time, and improved the robustness in the case of large differences between the training set and the test set.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗