Search NASA⌕ Search

SEARCH · Search NASA

Results for “Loss function”

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

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

At least 19 records

Entropy-Infused Deep Learning Loss Function for Capturing Extreme Values in Wind Power Forecasting

Extreme scenarios in wind power generation occur with higher frequency and larger magnitude in the recent years due to the ever-increasing extreme meteorological factors. Accurate forecasting of the occurrence of extreme values in wind power generation is of great concern to ensure reliable power system operation. Recently, deep learning models have surged in popularity for wind power forecasting, with the mean squared error (MSE) loss function being commonly used. However, the MSE loss function, being sensitive to extreme values, disproportionately penalizes larger errors, cannot adequately capture the extreme values present in wind energy data, and novel loss functions have seldom been tailored for wind power forecasting. To this end, in this paper, we introduce a novel loss function specifically crafted to capture extreme values in wind power forecasting. The experimental results with four fundamental deep learning methods on open source wind power dataset validate that the new loss function is efficient and superior in all cases compared to MSE in capturing extreme values while maintaining forecasting performance.

17 WIND ENERGY↗

Improved loss functions for machine-learned atomic potentials

Machine learning (ML) has become an invaluable tool across a wide array of domains in science as researchers find new ways to leverage its predictive power. This is especially true in chemistry, where ML is used to fit chemical properties or desirable attributes to the local structure of molecules and materials. In the pursuit of greater accuracy, it is relatively simple to increase the size or complexity of such models, although this often requires simultaneously seeking larger datasets in order to both fit and interpret the larger number of parameters. However, it is equally important to assess the quality and relative importance of the data and how these factors impact the training process. We, therefore, investigate the impact of using different loss functions for training neural network potentials (NNPs), as the loss function defines the error and parameter gradients used to train the NNP. In particular, we test the mean-squared error and Huber loss functions and, using insight from these functions, derive a new loss function based on the Asinh function, which yields significant improvement in the accuracy and generality of NNPs. We show that by discounting/minimizing errors and anomalies in the optimization process, both the Huber and Asinh loss functions improve the training of NNPs, leading to a final potential with a greater effective dimensionality.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Retrieving the energy-loss function from valence electron energy-loss spectrum: Separation of bulk-, surface-losses and Cherenkov radiation

With recent rapid advancement in electron microscopy instrumentation, in particular, bright electron sources and monochromators, valence electron energy-loss spectroscopy (VEELS) has become attractive for retrieving band structures, optical properties, dielectric functions and phonon information of materials. However, Cherenkov radiation and surface-loss contribution signi cantly alter ne structures of VEELS, even in simple semiconductors and insulators. This leads to the problem that dielectric function or bandgap structure of these materials cannot be determined correctly if these e ects are not removed. In this work we present a solution to this dilemma. We demonstrate that energy-loss function and real part of inverse complex dielectric function can be retrieved from raw data of VEELS. Based on the calculated example of Si, the limitation of our approach is discussed. We believe that our approach represents an improvement over previous procedures and has a broad prospect for applications.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Effect of neutral collision and radiative heat-loss function on self-gravitational instability of viscous thermally conducting partially-ionized plasma

The problem of thermal instability and gravitational instability is investigated for a partially ionized self-gravitating plasma which has connection in astrophysical condensations. We use normal mode analysis method in this problem. The general dispersion relation is derived using linearized perturbation equations of the problem. Effects of collisions with neutrals, radiative heat-loss function, viscosity, thermal conductivity and magnetic field strength, on the instability of the system are discussed. The conditions of instability are derived for a temperature-dependent and density-dependent heat-loss function with thermal conductivity. Numerical calculations have been performed to discuss the effect of various physical parameters on the growth rate of the gravitational instability. The temperature-dependent heat-loss function, thermal conductivity, viscosity, magnetic field and neutral collision have stabilizing effect, while density-dependent heat-loss function has a destabilizing effect on the growth rate of the gravitational instability. With the help of Routh-Hurwitz's criterion, the stability of the system is discussed.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Levenberg–Marquardt multi-classification using hinge loss function

Incorporating higher-order optimization functions, such as Levenberg-Marquardt (LM) have revealed better generalizable solutions for deep learning problems. However, these higher-order optimization functions suffer from very large processing time and training complexity especially as training datasets become large, such as in multi-view classification problems, where finding global optima is a very costly problem. To solve this issue, we develop a solution for LM-enabled classification with, to the best of knowledge first-time implementation of hinge loss, for multiview classification. Hinge loss allows the neural network to converge faster and perform better than other loss functions such as logistic or square loss rates. Here we prove our method by experimenting with various multiclass classification challenges of varying complexity and training data size. The empirical results show the training time and accuracy rates achieved, highlighting how our method outperforms in all cases, especially when training time is limited. Our paper presents important results in the relationship between optimization and loss functions and how these can impact deep learning problems.

97 MATHEMATICS AND COMPUTING↗

Precision measurement of the electron energy-loss function in tritium and deuterium gas for the KATRIN experiment

Abstract The KATRIN experiment is designed for a direct and model-independent determination of the effective electron anti-neutrino mass via a high-precision measurement of the tritium $$\upbeta $$ β -decay endpoint region with a sensitivity on $$m_\nu $$ m ν of 0.2 $$\hbox {eV}/\hbox {c}^2$$ eV / c 2 (90% CL). For this purpose, the $$\upbeta $$ β -electrons from a high-luminosity windowless gaseous tritium source traversing an electrostatic retarding spectrometer are counted to obtain an integral spectrum around the endpoint energy of 18.6 keV. A dominant systematic effect of the response of the experimental setup is the energy loss of $$\upbeta $$ β -electrons from elastic and inelastic scattering off tritium molecules within the source. We determined the energy-loss function in-situ with a pulsed angular-selective and monoenergetic photoelectron source at various tritium-source densities. The data was recorded in integral and differential modes; the latter was achieved by using a novel time-of-flight technique. We developed a semi-empirical parametrization for the energy-loss function for the scattering of 18.6-keV electrons from hydrogen isotopologs. This model was fit to measurement data with a 95% $$\hbox {T}_2$$ T 2 gas mixture at 30 K, as used in the first KATRIN neutrino-mass analyses, as well as a $$\hbox {D}_2$$ D 2 gas mixture of 96% purity used in KATRIN commissioning runs. The achieved precision on the energy-loss function has abated the corresponding uncertainty of $$\sigma (m_\nu ^2)< {{10}^{-2}}{\hbox {eV}^{2}}$$ σ ( m ν 2 ) < 10 - 2 eV 2 [1] in the KATRIN neutrino-mass measurement to a subdominant level.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Adaptive Quantum Generative Training using an Unbounded Loss Function

We propose a generative quantum learning algorithm using the Adaptive Derivative-Assembled Problem Tailored ansatz (ADAPT) framework in which the loss function to be minimized is the maximal quantum Rényi divergence of order two, an unbounded function that mitigates barren plateaus which inhibit training variational circuits. We benchmark this method against other state-of-the-art adaptive algorithms by learning random two-local thermal states. We perform numerical experiments of up to 12 qubits comparing our method learning algorithms that use linear objective functions and show that Rényi-ADAPT is capable of constructing shallow quantum circuits competitive with existing methods, while the gradients remain favorable resulting from the maximal Rényi divergence loss function.

quantum algorithms, quantum machine learning, quan↗

Accurately estimating and minimizing costs for the cellulosic biomass supply chain with statistical process control and the Taguchi Loss Function

This research focuses on the statistical evaluation of the feedstock attributes of the biomass supply chain and the estimation of attribute costs as a function of the feedstock variability. Challenges of using cellulosic feedstocks include the variability of feedstock quality (e.g., ash content and moisture content), which impacts the final cost of the manufactured product. Statistical Process Control (SPC), Taguchi Loss Function, and components of variance techniques were illustrated for quantifying cumulative variance in the biomass supply chain. Costs in the presence of cumulative variance were estimated for switchgrass (Panicum virgatum L.) and loblolly pine residues (Pinus taeda L.). Findings of the study indicated that additional costs from ash content variability in switchgrass increased the net cost by $\$19.15$ per dry tonne. Additional costs from densification due to particle size variation increased net cost by $\$11.59$ per dry tonne. Moisture content variation increased costs by $\$14.86$ per dry tonne. This would represent a 50 to 100% increase in costs due to variation based on a $\$60$ to $\$70$ per dry tonne manufactured product cost. This study illustrates that total costs may be considerably underestimated if the influence of variance for key factors in the supply chain and associated costs are not estimated.

59 BASIC BIOLOGICAL SCIENCES↗

Jensen–Shannon divergence based novel loss functions for Bayesian neural networks

Bayesian neural networks (BNNs) are state-of-the-art machine learning methods that can naturally regularize and systematically quantify uncertainties using their stochastic parameters. Kullback–Leibler (KL) divergence-based variational inference used in BNNs suffer from unstable optimization and challenges in approximating light-tailed posteriors due to the unbounded nature of the KL divergence. To resolve these issues, we formulate a novel loss function for BNNs based on a new modification to the generalized Jensen–Shannon (JS) divergence, which is bounded. In addition, we propose a Geometric JS divergence-based loss, which is computationally efficient since it can be evaluated analytically. We found that the JS divergence-based variational inference is intractable, and hence employed a constrained optimization framework to formulate these losses. Our theoretical analysis and empirical experiments on multiple regression and classification data sets suggest that the proposed losses perform better than the KL divergence-based loss, especially when the data sets are noisy or biased. Specifically, there are approximately 5% and 8% improvements in accuracy for a noise-added CIFAR-10 dataset and a regression dataset, respectively. There is about 13% reduction in false negative predictions of a biased histopathology dataset. Additionally, we quantify and compare the uncertainty metrics for the regression and classification tasks.

97 MATHEMATICS AND COMPUTING↗

Origin of the second peak in the mechanical loss function of amorphous silica

The thermal noise in amorphous oxides is the limiting factor for gravitational wave detectors and other high-precision optical devices. Through the fluctuation-dissipation theorem, the thermal noise is directly connected to the internal friction ($Q^{–1}$). Computational calculations of $Q^{–1}$ that use a two-level system (TLS) model have previously been performed for several coating materials, facilitating the search for coatings with lower thermal noise. However, they are based on a historical approximation made within the TLS model that treats the TLS distribution as uncorrelated, which has limited the predictive power of the model. In this paper, we show that this approximation limits the physical description of amorphous oxides using the TLS model and a fully correlated distribution must be used to calculate high-temperature behavior. Not only does using a correlated distribution improve the theoretical standing of the TLS model, calculations of $Q^{–1}$ using a fully correlated distribution reproduce and uncover the physical mechanisms of a second peak observed in measurements of ion-beam sputtered amorphous silica. We also explore the details of the thermal activation of TLSs and analyze the atomic transitions that contribute to $Q^{–1}$ in amorphous silica.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Excited State Electronic Properties of Sodium Iodide and Cesium Iodide

We compute from first principles the dielectric function, loss function, lifetime and scattering rate of quasiparticles due to electronic losses, and secondary particle spectrum due to plasmon decay in two scintillating alkali halides, sodium iodide and cesium iodide. Particular emphasis is placed on quasiparticles within several multiples of the band gap from the band edges. A theory for the decay spectra of plasmons and other electronic excitations in crystals is presented. Applications to Monte Carlo radiation transport codes are discussed.

Campbell, Luke W↗

Progressive Loss of Function in a Limb Enhancer during Snake Evolution

The evolution of body shape is thought to be tightly coupled to changes in regulatory sequences, but specific molecular events associated with major morphological transitions in vertebrates have remained elusive. In this paper, we identified snake-specific sequence changes within an otherwise highly conserved long-range limb enhancer of Sonic hedgehog (Shh). Transgenic mouse reporter assays revealed that the in vivo activity pattern of the enhancer is conserved across a wide range of vertebrates, including fish, but not in snakes. Genomic substitution of the mouse enhancer with its human or fish ortholog results in normal limb development. In contrast, replacement with snake orthologs caused severe limb reduction. Synthetic restoration of a single transcription factor binding site lost in the snake lineage reinstated full in vivo function to the snake enhancer. Our results demonstrate changes in a regulatory sequence associated with a major body plan transition and highlight the role of enhancers in morphological evolution.

59 BASIC BIOLOGICAL SCIENCES↗