Scaling Model Parameters Research Articles

Efficient Methods for Natural Language Processing: A Survey

AbstractRecent work in natural language processing (NLP) has yielded appealing results from scaling model parameters and training data; however, using only scale to improve performance means that resource consumption also grows. Such resources include data, time, storage, or energy, all of which are naturally limited and unevenly distributed. This motivates research into efficient methods that require fewer resources to achieve similar results. This survey synthesizes and relates current methods and findings in efficient NLP. We aim to provide both guidance for conducting NLP under limited resources, and point towards promising research directions for developing more efficient methods.

Optimized scaling of translational factors in oncology: from xenografts to RECIST

PurposeTumor growth inhibition (TGI) models are regularly used to quantify the PK–PD relationship between drug concentration and in vivo efficacy in oncology. These models are typically calibrated with data from xenograft mice and before being used for clinical predictions, translational methods have to be applied. Currently, such methods are commonly based on replacing model components or scaling of model parameters. However, difficulties remain in how to accurately account for inter-species differences. Therefore, more research must be done before xenograft data can fully be utilized to predict clinical response.MethodTo contribute to this research, we have calibrated TGI models to xenograft data for three drug combinations using the nonlinear mixed effects framework. The models were translated by replacing mice exposure with human exposure and used to make predictions of clinical response. Furthermore, in search of a better way of translating these models, we estimated an optimal way of scaling model parameters given the available clinical data.ResultsThe predictions were compared with clinical data and we found that clinical efficacy was overestimated. The estimated optimal scaling factors were similar to a standard allometric scaling exponent of − 0.25.ConclusionsWe believe that given more data, our methodology could contribute to increasing the translational capabilities of TGI models. More specifically, an appropriate translational method could be developed for drugs with the same mechanism of action, which would allow for all preclinical data to be leveraged for new drugs of the same class. This would ensure that fewer clinically inefficacious drugs are tested in clinical trials.

Psychometric Validation of the Athens Insomnia Scale Among Nurses: A Robust Approach Using Both Classical Theory and Rating Scale Model Parameters.

PurposeFew studies have investigated the validity of the Athens insomnia scale (AIS) using a robust approach of both classical theory and the rating scale model. Therefore, in this study, we investigated psychometric validation of the AIS using both of these approaches in nurses.MethodsNurses (n= 563, age= 33.2±7.1 years) working in health facilities in Saudi Arabia participated in a cross-sectional study. Participants completed the AIS, socio-demographics tool, and sleep health-related questions.ResultsConfirmatory factor analysis (CFA) favored a 2-factor structure with both comparative fit index (CFI), and incremental fit index (IFI) having values above 0.95. The 2-factor model had the lowest values of Akaike information criterion (AIC), root mean square error of approximation (RMSEA), χ2, and χ2/df. This 2-factor structure showed configural invariance (CFI more than 0.95, RMSEA less than 0.08, and Χ2/df less than 3), and metric, scalar, and strict invariance (based on Δ CFI ≤-0.01, and Δ RMSEA ≥ 0.015 criteria). No ceiling/floor effects were seen for the AIS total scores. Infit and outfit mean square values for all the items were within the acceptable range (<1.4, >0.6). The threshold estimates for each item were ordered as expected. Cronbach’s α for the AIS tool, factor-1 score, factor-2 score was 0.86, 0.82, and 0.72, respectively. AIS factor scores-1/2 were significantly associated with a habitual feeling of tiredness after usual night sleep (p<0.001), Impairment of daytime socio-occupational functioning (p<0.05), and with a feeling of daytime fatigue, irritability, and restlessness (p<0.05).ConclusionThe findings favor the validity of a 2-factor structure of the AIS with adequate item properties, convergent validity, and reliability in nurses.

THE INFLUENCE OF AUTOMATIC FLAP GATE ON EROSION IN TIDAL SWAMP AREA

This research aims to examine the problem of erosion and sedimentation in canals in swampy areas in South Sumatra, namely the P8-13S scheme of the canal (SPD) Telang I, Banyuasin regency with the influence of flap gate. The research method used is to perform simulations in the laboratory using a scale model and related parameters, such as velocity (v), depth of erosion (de), sedimentation (ds), water density (ρw), gravitation (g), time (t) and distance (x). Dimensional analysis to determine the related parameters that produce a dimensionless parameter relationship model is carried out using the Langhaar method. The result of the study was that the maximum relative erosion (de/t)max occurred at the canals estuary near the flap gate, which was 0.255 during simulation times of 5 minutes. While the maximum relative sedimentation (dd/t)max also occurred at the estuary of the secondary canals (SPD) of 0.05 for 5 minutes. So, the sediment transport process in the secondary canal (SPD) is generally influenced by flow velocity and the presence of a flap gate in the canal to regulate sediment transport.

Qualifying drug dosing regimens in pediatrics using Gaussian processes.

Drug development commonly studies an adult population first and then the pediatric population. The knowledge from the adult population is taken advantage of for the design of the pediatric trials. Adjusted drug doses for these are often derived from adult pharmacokinetic (PK) models which are extrapolated to patients with smaller body size. This extrapolation is based on scaling physiologic model parameters with a body size measure accounting for organ size differences. The inherent assumption is that children are merely small adults. However, children can be subject to additional effects such as organ maturation. These effects are not present in the adult population and are possibly overlooked at the design stage of the pediatric trials. It is thus crucial to qualify the extrapolation assumptions once the pediatric trial data are available. In this work, we propose a model based on a non-parametric regression method called Gaussian process (GP) to detect deviations from the made extrapolation assumptions. We introduce the theoretical background of this model and compare its performance to a parametric expansion of the adult model. The comparison includes simulations and a clinical study data example. The results show that the GP approach can reliably detect maturation trends from sparse pediatric data.

Research on Five-Loop Simplified Scaling Model for Dry-Type Air-Core Reactors With Inter-Turn Short Circuit Fault

In order to simplify the research method of the spatial magnetic field distribution (SMFD) of dry air-core reactors (DARs) in the case of inter-turn short circuit fault, the concept of simplified scaling model was put forward, and the optimal simplified scaling model for different inter-turn short circuit fault was calculated by using the least square method. Firstly, in order to simplify the calculation, 7 typical directions describing the SMFD of DARs were determined based on the similarity degree of magnetic induction intensity in each direction. Then, the simplified scaling model parameters corresponding to different short circuit positions and degrees were studied, and the variation rule was described by using the method of fitting function. Finally, the original model and the simplified scaling model of the DARs in the case of inter-turn short circuit were established respectively in the laboratory. The results show that the optimal structure relative height obtained by experiment is 30% and 20% respectively, which is the closest to the parameters obtained by computer simulation, i.e.,28% and 20.3%.

Efficient parameterization of large-scale dynamic models based on relative measurements.

Mechanistic models of biochemical reaction networks facilitate the quantitative understanding of biological processes and the integration of heterogeneous datasets. However, some biological processes require the consideration of comprehensive reaction networks and therefore large-scale models. Parameter estimation for such models poses great challenges, in particular when the data are on a relative scale. Here, we propose a novel hierarchical approach combining (i) the efficient analytic evaluation of optimal scaling, offset and error model parameters with (ii) the scalable evaluation of objective function gradients using adjoint sensitivity analysis. We evaluate the properties of the methods by parameterizing a pan-cancer ordinary differential equation model (>1000 state variables, >4000 parameters) using relative protein, phosphoprotein and viability measurements. The hierarchical formulation improves optimizer performance considerably. Furthermore, we show that this approach allows estimating error model parameters with negligible computational overhead when no experimental estimates are available, providing an unbiased way to weight heterogeneous data. Overall, our hierarchical formulation is applicable to a wide range of models, and allows for the efficient parameterization of large-scale models based on heterogeneous relative measurements. Supplementary code and data are available online at http://doi.org/10.5281/zenodo.3254429 and http://doi.org/10.5281/zenodo.3254441. Supplementary data are available at Bioinformatics online.

Correlation Analysis of High-Resolution Particle Image Velocimetry Data of Screeching Jets

The acoustic analogy provides a general framework for predicting broadband jet noise. The accuracy of the noise predictions is strongly dependent on the second- and fourth-order integral time and length scales of the turbulence quantities in the jet. Two low-order models for the second- and fourth-order integral length scales are examined. The low-order models are defined by locally isotropic scales estimated from two-dimensional particle image velocimetry measurements. These measurements are of screeching underexpanded unheated round jets issuing from a purely converging nozzle at conditions that correspond to ideally expanded Mach numbers of 1.45 and 1.59. The jets are dominated by the helical (C) instability screech mode, which is associated with large-scale coherent periodic fluctuations. These fluctuations are filtered using a proper orthogonal decomposition method to assess low-order models that approximate the length scales associated with the broadband noise mechanisms. The length scale model parameters are shown to be insensitive for the two Mach numbers considered. The root-mean-square error associated with the low-order models indicates that either is sufficient for approximating the integral length scales required to model equivalent sources of broadband jet noise.

Interface Development of Ship Indoor Lighting

In this research study, a lighting interface was designed to draw lighting wiring diagrams of ships using a Visual Basic for Applications (VBA) program. By means of the utilization factor method for calculating the lamp number in the cabin, the luminaires and wiring are drawn with an AutoCAD drawing. From the design concept and manufacturing cost, the VBA program can build a construction layout, indoor lighting systems drawing, description of the luminaires, and cost estimation. Illumination experiments can calibrate the lamp number and illumination correctness; last, we use ant colony algorithms to calculate the optimized path and reduce the costs. 1. Introduction Light is a basic need of people in their daily lives, and it can affect people's physical and psychological behavior. Bellia et al. (2011) and Flynn et al. (1973) explained light sources with visual and nonvisual effects in terms of different spectral power, and the traditional method of lighting design uses the lowest illumination and lighting position to achieve an effective average illuminance of the environment. In addition to considering the lowest illumination, the visible and nonvisual aspects of lighting design must be estimated by designers. In general, lighting design is affected by daylight, and the daylight is from the sun and sky; the intensity of light in the sunlight is reduced in the atmosphere by the irradiation direction problem, and the sky's illumination is affected by the sun's radiation (Elvegård &amp; Sjöstedt 1940). In recent years, there have been many lighting designs using computer simulations that account for the effects of sunlight. For example, Ng et al. (2001) used simulation software for lighting to assist the interior lighting design of buildings after recording the sunshine data, which allowed designers to more efficiently control the building's lighting conditions. Li and Tsang (2005) simulated indoor corridor illumination; the simulation results would be detected by light sensors, and the compared results were used to predict the consumption of lighting energy and to achieve energy savings. Yun and Kim (2013) accurately predicted the lighting energy consumed by the integrated simulation method, setting the scale model parameters that are required for simulation, through the split-flux, radiosity, and Daysin daylighting algorithms; the simulation results were considered for comparing the lighting of the energy consumption. Díaz-Vilariño et al. (2014) used simulation software to automatically generate a 3D geometrical model that can import lighting simulation software to perform an illumination analysis. Jia et al. (2014) integrated a dynamic environment simulation and illumination algorithms to provide quantitative visualization for lighting and position, to simulate and evaluate the impact of the lighting system.

Grid size effects analysis and hydrological similarity of surface runoff in flatland basins

ABSTRACTThe effects of grid-size modification on the derived topographic attributes are analysed and a procedure for scaling model parameters and similarity assessment between flow variables is proposed. Hydrological simulations are performed with a physically-based and spatially-distributed quasi-2D mathematical model. The scaled model parameters are the effective roughness coefficient associated with overland flow (nov) and the transverse slope in the cell (TSC). To scale the selected parameters, the criterion of equilibrium storage conservation between the different grid sizes is applied. Three basins of the central-east region of Argentina are modelled. The spatial variability of basin geomorphology is quantified using the entropy concept. The simulation results show that when grid size is increased, to obtain similar hydrological responses it is necessary to increase the nov or to reduce the TSC. In terms of similarity, the best results are achieved when TSC is scaled, particularly when water depths are considered.

Model-based registration of preprocedure MR and intraprocedure US of the lumbar spine

Epidural and spinal needle insertions, as well as facet joint denervation and injections are widely performed procedures on the lumbar spine for delivering anesthesia and analgesia. Ultrasound (US)-based approaches have gained popularity for accurate needle placement, as they use a non-ionizing, inexpensive and accessible modality for guiding these procedures. However, due to the inherent difficulties in interpreting spinal US, they yet to become the clinical standard-of-care. A novel statistical shape[Formula: see text]pose[Formula: see text]scale (s[Formula: see text]p[Formula: see text]s) model of the lumbar spine is jointly registered to preoperative magnetic resonance (MR) and US images. An instance of the model is created for each modality. The shape and scale model parameters are jointly computed, while the pose parameters are estimated separately for each modality. The proposed method is successfully applied to nine pairs of preoperative clinical MR volumes and their corresponding US images. The results are assessed using the target registration error (TRE) metric in both MR and US domains. The s[Formula: see text]p[Formula: see text]s model in the proposed joint registration framework results in a mean TRE of 2.62 and 4.20mm for MR and US images, respectively, on different landmarks. The joint framework benefits from the complementary features in both modalities, leading to significantly smaller TREs compared to a model-to-US registration approach. The s[Formula: see text]p[Formula: see text]s model also outperforms our previous shape[Formula: see text]pose model of the lumbar spine, as separating scale from pose allows to better capture pose and guarantees equally-sized vertebrae in both modalities. Furthermore, the simultaneous visualization of the patient-specific models on the MR and US domains makes it possible for clinicians to better evaluate the local registration accuracy.

Uncertainty quantification in LES of channel flow

SummaryIn this paper, we present a Bayesian framework for estimating joint densities for large eddy simulation (LES) sub‐grid scale model parameters based on canonical forced isotropic turbulence direct numerical simulation (DNS) data. The framework accounts for noise in the independent variables, and we present alternative formulations for accounting for discrepancies between model and data. To generate probability densities for flow characteristics, posterior densities for sub‐grid scale model parameters are propagated forward through LES of channel flow and compared with DNS data. Synthesis of the calibration and prediction results demonstrates that model parameters have an explicit filter width dependence and are highly correlated. Discrepancies between DNS and calibrated LES results point to additional model form inadequacies that need to be accounted for. Copyright © 2016 John Wiley &amp; Sons, Ltd.

Optimization of a fully air-swept dry grinding cement raw meal ball mill closed circuit capacity with the aid of simulation

Production capacity of a fully air-swept industrial scale two-compartment KHD Humboldt Wedag® cement ball mill was optimized with the aid of simulation. It was proposed to operate the mill as a single compartment by eliminating the pre-drying compartment. In this respect, grinding performance of the air-swept ball mill was evaluated and modelled as a perfectly mixed single tank using the perfect mixing ball mill modelling approach (Whiten, 1974). Static separator was modelled by efficiency curve model (Whiten, 1966). The empirical breakage function required in the estimation of average specific breakage rates was measured by drop-weight technique. The full scale model parameters were used to simulate the raw meal mill grinding circuit with the aid of JKSimMet Steady State Mineral Processing Simulator. Simulation results indicated 23% production capacity increase in cement throughput in case the pre-drying compartment was used in grinding.

Interspecies Pharmacokinetic Modeling of Subcutaneous Absorption of Rituximab in Mice and Rats

To investigate the effect of dose level and anatomical site of injection on the pharmacokinetics of rituximab in mice, and to evaluate the utility of a pharmacokinetic model for describing interspecies differences in subcutaneous absorption between mice and rats. Rituximab serum concentrations were measured following intravenous and subcutaneous administration at the back and abdomen of mice. Several approaches were compared for scaling model parameters from estimated values in rats. The bioavailability of rituximab following subcutaneous injection was inversely related to the dose level and was dependent on the site of injection in mice. The overall rate of absorption was faster in mice as compared to rats. Subcutaneous absorption profiles were well described using the proposed structural model, in which the total receptor concentration, the affinity of rituximab to the receptor, and the degradation rate constant were assumed to be species independent. Subcutaneous absorption processes show similar trends in rats and mice, although the magnitude differs between species. A mathematical model that combines the absorption of free and bound antibody with presystemic degradation successfully captured rituximab pharmacokinetics in both species, and approaches for sharing and scaling parameters between species were identified.

Scale properties of the seismic wavefield perspectives for full-waveform matching

Starting from the nondimensionalization of equations of motion we partition the set of the velocity models in equivalence classes, such that the full waveform of an element in a given class can be calculated from the full waveform of any other element in the same class by scaling model parameters. We give a formal derivation of the seismic wavefield scale properties and we prove their capability through the use of numerical examples. Besides this, we introduce how the scale properties can be used to save computational time in full waveform modeling and inversion. In forward modeling we can use them for the calculation of the full waveform of any model in the same equivalence class of a model whose full waveform has been previously calculated. In full waveform inversion, scale properties can be used for full waveform matching: Given an experimental seismogram and a synthetic one, we can choose, in the same class of the synthetic model, another element whose waveform is closer to the experimental one.

Multifractality, imperfect scaling and hydrological properties of rainfall time series simulated by continuous universal multifractal and discrete random cascade models

Abstract. Discrete multiplicative random cascade (MRC) models were extensively studied and applied to disaggregate rainfall data, thanks to their formal simplicity and the small number of involved parameters. Focusing on temporal disaggregation, the rationale of these models is based on multiplying the value assumed by a physical attribute (e.g., rainfall intensity) at a given time scale L, by a suitable number b of random weights, to obtain b attribute values corresponding to statistically plausible observations at a smaller L/b time resolution. In the original formulation of the MRC models, the random weights were assumed to be independent and identically distributed. However, for several studies this hypothesis did not appear to be realistic for the observed rainfall series as the distribution of the weights was shown to depend on the space-time scale and rainfall intensity. Since these findings contrast with the scale invariance assumption behind the MRC models and impact on the applicability of these models, it is worth studying their nature. This study explores the possible presence of dependence of the parameters of two discrete MRC models on rainfall intensity and time scale, by analyzing point rainfall series with 5-min time resolution. Taking into account a discrete microcanonical (MC) model based on beta distribution and a discrete canonical beta-logstable (BLS), the analysis points out that the relations between the parameters and rainfall intensity across the time scales are detectable and can be modeled by a set of simple functions accounting for the parameter-rainfall intensity relationship, and another set describing the link between the parameters and the time scale. Therefore, MC and BLS models were modified to explicitly account for these relationships and compared with the continuous in scale universal multifractal (CUM) model, which is used as a physically based benchmark model. Monte Carlo simulations point out that the dependence of MC and BLS parameters on rainfall intensity and cascade scales can be recognized also in CUM series, meaning that these relations cannot be considered as a definitive sign of departure from multifractality. Even though the modified MC model is not properly a scaling model (parameters depend on rainfall intensity and scale), it reproduces the empirical traces of the moments and moment exponent function as effective as the CUM model. Moreover, the MC model is able to reproduce some rainfall properties of hydrological interest, such as the distribution of event rainfall amount, wet/dry spell length, and the autocorrelation function, better than its competitors owing to its strong, albeit unrealistic, conservative nature. Therefore, even though the CUM model represents the most parsimonious and the only physically/theoretically consistent model, results provided by MC model motivate, to some extent, the interest recognized in the literature for this type of discrete models.

Study of hydrogen auto-ignition in a turbulent air co-flow using a Large Eddy Simulation approach

Large Eddy Simulation (LES) is applied to the auto-ignition of an hydrogen jet issuing into a turbulent co-flowing air stream. A 19 step, 9 species detailed mechanism is used for modelling the chemical reactions. The influence of sub-grid fluctuations is accounted for by a sub-grid joint probability density function (PDF) for the reactive scalars. A Eulerian Stochastic Field method is used to solve the modelled form of the PDF transport equation. The model is able to reproduce ignition lengths and different regimes observed experimentally without adjustment of the sub-grid scale model parameters.

Spatial scaling properties of annual streamflow in the United States

The spatial scaling properties of annual average streamflow is examined using records from 1 433 river basins across the continental United States. The log-linear relationship ln(E[Qr i]) = a + br ln(Ai) is representative throughout the United States, where E[Qr i] represents the expectation of the rth moment of annual streamflow at site i, and Ai represents drainage area. The scaling model parameters ar and br follow nearly perfect linear relationships ar = rα and br = rβ throughout the continental United States. We conclude that the probability distribution of annual streamflow follows simple scaling relationships in all regions of the United States. In temperate regions where climate is relatively homogeneous, scale alone describes most of the variability in the moments of annual streamflow. In the more climatically heterogeneous regions, such as in the Upper Colorado and Missouri river basins, scale alone is a poor predictor of the moments of annual flow.

Measuring gauge-mediated supersymmetry breaking parameters at a 500 GeV $e^+e^-$ linear collider

We consider the phenomenology of a class of gauge-mediated supersymmetry (SUSY) breaking (GMSB) models at a e+e- Linear Collider (LC) with c.o.m. energy up to 500 GeV. In particular, we refer to a high-luminosity (L ~ 3 x 10^34 cm^-2 s^-1) machine, and use detailed simulation tools for a proposed detector. Among the GMSB-model building options, we define a simple framework and outline its predictions at the LC, under the assumption that no SUSY signal is detected at LEP or Tevatron. Our focus is on the case where a neutralino (N1) is the next-to-lightest SUSY particle (NLSP), for which we determine the relevant regions of the GMSB parameter space. Many observables are calculated and discussed, including production cross sections, NLSP decay widths, branching ratios and distributions, for dominant and rare channels. We sketch how to extract the messenger and electroweak scale model parameters from a spectrum measured via, e.g. threshold-scanning techniques. Several experimental methods to measure the NLSP mass and lifetime are proposed and simulated in detail. We show that these methods can cover most of the lifetime range allowed by perturbativity requirements and suggested by cosmology in GMSB models. Also, they are relevant for any general low-energy SUSY breaking scenario. Values of c*tau_N1 as short as 10's of microns and as long as 10's of metres can be measured with errors at the level of 10% or better after one year of LC running with high luminosity. We discuss how to determine a narrow range (<~ 5%) for the fundamental SUSY breaking scale sqrt(F), based on the measured m_N1, c*tau_N1. Finally, we suggest how to optimise the LC detector performance for this purpose.

Optimal estimation of gradient damage parameters from localization phenomena in quasi‐brittle materials

Non-local and gradient enhanced damage models are able to properly model localization phenomena in quasi-brittle materials. By the introduction of an internal length scale they avoid mesh-size sensitivity in finite element calculations and are capable of describing size effects. To experimentally determine the internal length scale and other damage model parameters, indirect identification methods have to be used. The method presented is based on the Markov estimation procedure, which also permits to evaluate the accuracy and the well-posedness of the identification problem, characterized by the uniqueness and identifiability of the estimated values. This information offers the possibility to compare and optimize the designs of laboratory tests in order to minimize the effect of errors on the estimated values. The indirect determination of the gradient damage parameters based on global response measurements far from the crack on one-size specimens is sensitive to the problem of ill-posedness due the high correlation between the gradient damage parameters, which dominate the localization process. The size effect method can be used as an accurate identification method for the gradient damage parameters, when information on the tensile behaviour is included and a sufficient size range is considered. Local displacement analysis close to the crack provides sufficient information with respect to the identification problem independent of the size of the specimen and gives the most accurate parameter estimations. Copyright © 1999 John Wiley & Sons, Ltd.