Discrepant Estimates Research Articles

Sex determination accuracy using the human sacrum in a Central European sample.

The sacrum is often used for sex estimation of human remains. However, reported sexual classification accuracies based on the sacrum produce highly discrepant estimates ranging from 54% to 98% depending on the method and reference sample. Here, we therefore aim to evaluate the sacrum's efficacy in differentiating males from females by comparing three different approaches in a homogeneous Central European sample (n = 58, 27 males and 31 females) obtained from the 19th century Weisbach collection. Specifically, we investigated the sacrum by 1) a qualitative visual inspection, 2) traditional linear metrics, namely, sacral width, corpus width, and the associated corporo-basal index, and 3) geometric morphometrics (GM) using a 3D configuration of 100 landmarks and semilandmarks. Classification accuracies for the qualitative approach ranged from 69% to 81%. The investigated quantitative methods based on linear dimensions led to comparable classification accuracies of 62% for sacral width and 78% for the corporo-basal index. However, absolute corpus width had a stronger discriminative power (86%), similar to the PC scores from shape space (87%). Unexpectedly, the GM approach, which considered the shape of the entire sacrum, did not classify better than linear variables. This was exceeded only when sacrum shape was combined with corpus width or another measure of size, like the natural logarithm of the centroid size, yielding a classification accuracy of 95%. The male and female group means differed mainly in the general height-to-width relationship and corpus-to-alae proportions, corroborating patterns previously described in the literature. Our results suggest that the sacrum is markedly less effective than the hipbone for sex determination, which is at odds with its central position within the pelvic girdle and thus its supposed obstetric relevance.

SPICE: Semantic Pseudo-Labeling for Image Clustering.

The similarity among samples and the discrepancy among clusters are two crucial aspects of image clustering. However, current deep clustering methods suffer from inaccurate estimation of either feature similarity or semantic discrepancy. In this paper, we present a Semantic Pseudo-labeling-based Image ClustEring (SPICE) framework, which divides the clustering network into a feature model for measuring the instance-level similarity and a clustering head for identifying the cluster-level discrepancy. We design two semantics-aware pseudo-labeling algorithms, prototype pseudo-labeling and reliable pseudo-labeling, which enable accurate and reliable self-supervision over clustering. Without using any ground-truth label, we optimize the clustering network in three stages: 1) train the feature model through contrastive learning to measure the instance similarity; 2) train the clustering head with the prototype pseudo-labeling algorithm to identify cluster semantics; and 3) jointly train the feature model and clustering head with the reliable pseudo-labeling algorithm to improve the clustering performance. Extensive experimental results demonstrate that SPICE achieves significant improvements (~10%) over existing methods and establishes the new state-of-the-art clustering results on six balanced benchmark datasets in terms of three popular metrics. Importantly, SPICE significantly reduces the gap between unsupervised and fully-supervised classification; e.g. there is only 2% (91.8% vs 93.8%) accuracy difference on CIFAR-10. Our code is made publicly available at https://github.com/niuchuangnn/SPICE.

A sharp estimate of the discrepancy of a concatenation sequence of inversive pseudorandom numbers with consecutive primes

We consider an equidistributed concatenation sequence of pseudorandom rational numbers generated from the primes by an inversive congruential method. In particular, we determine the sharp convergence rate for the star discrepancy of said sequence. Our arguments are based on well-known discrepancy estimates for inversive congruential pseudorandom numbers together with asymptotic formulae involving prime numbers.

High variability in results and methodological quality among overlapping systematic reviews on the same topics in surgery: a meta-epidemiological study.

Redundant publication of systematic reviews and meta-analyses (SRs/MAs) on the same topic presents an increasing burden for clinicians. The aim of this study was to describe variabilities in effect size and methodological quality of overlapping surgery-related SRs/MAs and to investigate factors associated with their postpublication citations. PubMed/MEDLINE was searched to identify SRs/MAs of RCTs on thoracoabdominal surgeries published in 2015. Previous SRs/MAs on the same topics published within the preceding 5 years (2011-2015) were identified and 5-year citation counts (through to 2020) were evaluated. Discrepancies in pooled effect sizes and their methodological quality using A Measurement Tool to Assess Systematic Reviews (AMSTAR) among overlapping SRs/MAs were assessed. The SR/MA-level factors associated with 5-year citation counts were explored, using a mixed-effects regression model with a random intercept for surgical topics. A total of 57 surgery-related SRs/MAs (48 topics) published in 2015 were identified, and 146 SRs/MAs had overlapping publications on 29 topics (60.4 per cent of all topics) in the preceding 5 years. There was considerable variability in methodological quality of SRs/MAs and coverage probability for relevant RCTs, resulting in discrepant effect size estimates for the same topic. High quality (AMSTAR score 8-11) was independently associated with higher 5-year citation counts (coefficient = 32.82; 95 per cent c.i. 15.63 to 50.02; P < 0.001). Overlapping SRs/MAs with high variability in results and methodological quality were common in surgery. A high-quality SR/MA score was an independent predictor of more frequent citations. Researchers and journal editors should concentrate their efforts on limiting publications to higher-quality reviews.

Online geometry monitoring during directed energy deposition additive manufacturing using laser line scanning

Additive manufacturing (AM) is a powerful and promising manufacturing technology. Advances in AM have led to increasing demands for its applications; however, controlling the geometry of deposited objects during the AM process remains a major challenge. Geometry control and compliance to a target geometry are some of the key concerns for AM, especially for a prototype requiring high precision in its geometry. In this study, an online geometry estimation methodology for continuous monitoring during the directed energy deposition (DED) process was developed using a laser line scanner. Our proposed methodology comprises (1) real-time scanning of each deposited track's profile, (2) online extraction of the track's geometry, and (3) online plotting and comparison of the as-designed and as-built models. In the methodology, data analysis following real-time scanning, such as geometry discrepancy estimation and online plotting of the as-built model, is attained for both single- and multi-layer objects. The effectiveness of our developed methodology is examined by comparing the profiles of the single- and multi-layer objects estimated during the DED process with the reference profiles obtained via laser line scanning and microscopy after the completion of the DED process.

A Low Discrepancy Sequence on Graphs

Many applications such as election forecasting, environmental monitoring, health policy, and graph based machine learning require taking expectation of functions defined on the vertices of a graph. We describe a construction of a sampling scheme analogous to the so called Leja points in complex potential theory that can be proved to give low discrepancy estimates for the approximation of the expected value by the impirical expected value based on these points. In contrast to classical potential theory where the kernel is fixed and the equilibrium distribution depends upon the kernel, we fix a probability distribution and construct a kernel (which represents the graph structure) for which the equilibrium distribution is the given probability distribution. Our estimates do not depend upon the size of the graph.

Harmonization status of procalcitonin measurements: what do comparison studies and EQA schemes tell us?

Sepsis represents a global health priority because of its high mortality and morbidity. The key to improving prognosis remains an early diagnosis to initiate appropriate antibiotic treatment. Procalcitonin (PCT) is a recognized biomarker for the early indication of bacterial infections and a valuable tool to guide and individualize antibiotic treatment. To meet the increasing demand for PCT testing, numerous PCT immunoassays have been developed and commercialized, but results have been questioned. Many comparison studies have been carried out to evaluate analytical performance and comparability of results provided by the different commercially available immunoassays for PCT, but results are conflicting. External Quality Assessment Schemes (EQAS) for PCT constitute another way to evaluate results comparability. However, when making this comparison, it must be taken into account that the variety of EQA materials consist of different matrices, the commutability of which has not yet been investigated. The present study gathers results from all published comparison studies and results from 137 EQAS surveys to describe the current state-of-the-art harmonization of PCT results. Comparison studies globally highlight a significant variability of measurement results that nonetheless seem to have a moderate impact on medical decision-making. For their part, EQAS for PCT provides highly discrepant estimates of the interlaboratory CV. Due to differences in commutability of the EQA materials, the results from different peer groups could not be compared. To improve the informative value of the EQA data, the existing limitations such as non-harmonized conditions and suboptimal and/or unknown commutability of the EQA materials have to be overcome. The study highlights the need for commutable reference materials that could be used to properly evaluate result comparability and possibly standardize calibration, if necessary. Such an initiative would further improve the safe use of PCT in clinical routine.

Uniform Distribution of the Weighted Sum-of-Digits Functions

Abstract The higher-dimensional generalization of the weighted q-adic sum-of-digits functions sq,γ (n), n =0, 1, 2,..., covers several important cases of sequences investigated in the theory of uniformly distributed sequences, e.g., d-dimensional van der Corput-Halton or d-dimensional Kronecker sequences. We prove a necessary and sufficient condition for the higher-dimensional weighted q-adic sum-of-digits functions to be uniformly distributed modulo one in terms of a trigonometric product. As applications of our condition we prove some upper estimates of the extreme discrepancies of such sequences, and that the existence of distribution function g(x)= x implies the uniform distribution modulo one of the weighted q-adic sum-of-digits function sq,γ (n), n = 0, 1, 2,... We also prove the uniform distribution modulo one of related sequences h 1 sq, γ (n)+h 2 sq,γ (n +1), where h 1 and h 2 are integers such that h 1 + h 2 ≠ 0 and that the akin two-dimensional sequence sq,γ (n), sq,γ (n +1) cannot be uniformly distributed modulo one if q ≥ 3. The properties of the two-dimensional sequence sq,γ (n),s q,γ (n +1), n =0, 1, 2,..., will be instrumental in the proofs of the final section, where we show how the growth properties of the sequence of weights influence the distribution of values of the weighted sum-of-digits function which in turn imply a new property of the van der Corput sequence.

The Black Hole Spin in GRS 1915+105, Revisited

Abstract We estimate the black hole spin parameter in GRS 1915+105 using the continuum-fitting method with revised mass and inclination constraints based on the very long baseline interferometric parallax measurement of the distance to this source. We fit Rossi X-ray Timing Explorer observations selected to be accretion-disk-dominated spectral states as described in McClintock et al. (2006) and Middleton et al. (2006), which previously gave discrepant spin estimates with this method. We find that, using the new system parameters, the spin in both data sets increased, providing a best-fit spin of a * = 0.86 for the Middleton et al. data and a poor fit for the McClintock et al. data set, which becomes pegged at the BHSPEC model limit of a * = 0.99. We explore the impact of the uncertainties in the system parameters, showing that the best-fit spin ranges from a * = 0.4 to 0.99 for the Middleton et al. data set and allows reasonable fits to the McClintock et al. data set with near-maximal spin for system distances greater than ∼10 kpc. We discuss the uncertainties and implications of these estimates.

GEOPOLIS software tools – Groundwater flow modeling at deep repository of liquid radioactive waste at Severny test site

This article discusses a three-dimensional groundwater flow model of a deep disposal facility at Severny test site. The three-dimensional model is a part of the certified software GEOPOLIS, based on the hydrogeological code GeRa (Geomigration of Radionuclides) serving as the calculation engine. This study describes the hydrogeological patterning of the groundwater flow model, as well as the results of calibration and verification of the model water heads with respect to the data of monitoring for more than 40 years of the deep repository exploitation. The article begins with a brief overview of the previously developed hydrogeological models of this object and continues with a description of the geological structure of the territory, and with a substantiation of the boundaries and parameters of the model. The results of groundwater flow modeling, model calibration, verification and estimation of discrepancy between the model results and monitoring data are shown. The comparison of the modeled and observed water heads in the stationary conditions (before the start of injection) and during operation of the deep repository allows making conclusion on the quality of calibration.

COMPARATIVE ANALYSIS OF METHODS FOR DETERMINATION OF SUPPLY FLOWS

Problem. The capacity of signalized intersections is determined using the concept of "saturation flow". It is the main characteristic in the process of their analysis, design and calculation. The work is devoted to solving the scientific-applied problem of increasing the efficiency of the signalized intersection at one level by choosing a rational method of determining saturation flows for these characteristics of traffic flows, geometric parameters of the intersection and infrastructure in its area. The subject of the study is the influence of the main characteristics of the traffic flow, parameters and infrastructure of the intersection on the saturation flow value. Goal. The aim of the work is the analysis of the influence of the main characteristics of the traffic flow, parameters and infrastructure of the intersection on the saturation flow value. Methodology. The calculation of the control regime at the intersection is based on determining the relation between the traffic intensity of vehicles at the intersection and the saturation flow value, which changes under the influence of a number of factors. Analytical calculation of the saturation flow value is based on using of correction factors that take into account the influence of these factors. Results. The methods of determination of saturation flows are analyzed out in the work; performed a comparative analysis of the definition of ideal saturation flows by different methods; a comparative analysis of the determination of the saturation flow correction coefficients for the parameters that have an effect on it were performed; the estimation of discrepancy at calculations of coefficients of correction of saturation flows by various methods is done; the list of coefficients which need to be considered at definition of their influence on a saturation flow is refined. Originality. In oppose to classical and other approaches to determining the saturation flow, the method recommended by the American "Highway Capacity Manual" takes into account more parameters of influence. It is proposed to use this method for the saturation flow calculation, but with an updated list of coefficients that must be taken into account in the calculations. Practical value. The results of the work can be used in the practice of improving the efficiency of signalized intersections in sections of the urban road network, as well as in further research on this topic.

On the triple correlations of fractional parts of

AbstractFor fixed$\alpha \in [0,1]$, consider the set$S_{\alpha ,N}$of dilated squares$\alpha , 4\alpha , 9\alpha , \dots , N^2\alpha \, $modulo$1$. Rudnick and Sarnak conjectured that, for Lebesgue, almost all such$\alpha $the gap-distribution of$S_{\alpha ,N}$is consistent with the Poisson model (in the limit asNtends to infinity). In this paper, we prove a new estimate for the triple correlations associated with this problem, establishing an asymptotic expression for the third moment of the number of elements of$S_{\alpha ,N}$in a random interval of length$L/N$, provided that$L&gt; N^{1/4+\varepsilon }$. The threshold of$\tfrac {1}{4}$is substantially smaller than the threshold of$\tfrac {1}{2}$(which is the threshold that would be given by a naïve discrepancy estimate).Unlike the theory of pair correlations, rather little is known about triple correlations of the dilations$(\alpha a_n \, \text {mod } 1)_{n=1}^{\infty } $for a nonlacunary sequence$(a_n)_{n=1}^{\infty } $of increasing integers. This is partially due to the fact that the second moment of the triple correlation function is difficult to control, and thus standard techniques involving variance bounds are not applicable. We circumvent this impasse by using an argument inspired by works of Rudnick, Sarnak, and Zaharescu, and Heath-Brown, which connects the triple correlation function to some modular counting problems.In Appendix B, we comment on the relationship between discrepancy and correlation functions, answering a question of Steinerberger.

Breast cancer screening and overdiagnosis.

Overdiagnosis is a harmful consequence of screening which is particularly challenging to estimate. An unbiased setting to measure overdiagnosis in breast cancer screening requires comparative data from a screened and an unscreened cohort for at least 30 years. Such randomised data will not become available, leaving us with observational data over shorter time periods and outcomes of modelling. This collaborative effort of the International Cancer Screening Network quantified the variation in estimated breast cancer overdiagnosis in organised programmes with evaluation of both observed and simulated data, and presented examples of how modelling can provide additional insights. Reliable observational data, analysed with study design accounting for methodological pitfalls, and modelling studies with different approaches, indicate that overdiagnosis accounts for less than 10% of invasive breast cancer cases in a screening target population of women aged 50 to 69. Estimates above this level are likely to derive from inaccuracies in study design. The widely discrepant estimates of overdiagnosis reported from observational data could substantially be reduced by use of a cohort study design with at least 10 years of follow-up after screening stops. In contexts where concomitant opportunistic screening or gradual implementation of screening occurs, and data on valid comparison groups are not readily available, modelling of screening intervention becomes an advantageous option to obtain reliable estimates of breast cancer overdiagnosis.

Discrepancy of Minimal Riesz Energy Points

We find upper bounds for the spherical cap discrepancy of the set of minimizers of the Riesz s-energy on the sphere \(\mathbb S^d.\) Our results are based on bounds for a Sobolev discrepancy introduced by Thomas Wolff in an unpublished manuscript where estimates for the spherical cap discrepancy of the logarithmic energy minimizers in \(\mathbb S^2\) were obtained. Our result improves previously known bounds for \(0\le s<2\) and \(s\ne 1\) in \(\mathbb S^2,\) where \(s=0\) is Wolff’s result, and for \(d-t_0<s<d\) with \(t_0\approx 2.5\) when \(d\ge 3\) and \(s\ne d-1.\)

Bounds for discrepancies in the Hamming space

We derive bounds for the ball Lp-discrepancies in the Hamming space for 0<p<∞ and p=∞. Sharp estimates of discrepancies have been obtained for many spaces such as the Euclidean spheres and more general compact Riemannian manifolds. In the present paper, we show that the behavior of discrepancies in the Hamming space differs fundamentally because the volume of the ball in this space depends on its radius exponentially while such a dependence for the Riemannian manifolds is polynomial.

Bayesian Neural Networks with Maximum Mean Discrepancy regularization

Bayesian Neural Networks (BNNs) are trained to optimize an entire distribution over their weights instead of a single set, having significant advantages in terms of, e.g., interpretability, multi-task learning, and calibration. Because of the intractability of the resulting optimization problem, most BNNs are either sampled through Monte Carlo methods, or trained by minimizing a suitable Evidence Lower BOund (ELBO) on a variational approximation. In this paper, we propose an optimized version of the latter, wherein we replace the Kullback–Leibler divergence in the ELBO term with a Maximum Mean Discrepancy (MMD) estimator, inspired by recent work in variational inference. After motivating our proposal based on the properties of the MMD term, we proceed to show a number of empirical advantages of the proposed formulation over the state-of-the-art. In particular, our BNNs achieve higher accuracy on multiple benchmarks, including several image classification tasks. In addition, they are more robust to the selection of a prior over the weights, and they are better calibrated. As a second contribution, we provide a new formulation for estimating the uncertainty on a given prediction, showing it performs in a more robust fashion against adversarial attacks and the injection of noise over their inputs, compared to more classical criteria such as the differential entropy.

Adjusting a torsional vibration damper model with physics-informed neural networks

In this work, we implement a framework for adjusting the outputs of a torsional vibration damper (TVD) model to experimental data using physics-informed neural networks. TVDs are devices used to passively control vibration; and here are commonly modeled through reduced-order physics. Within the TVD model, the material properties of the viscoelastic rubber used in the device are characterized through previously performed coupon tests. Even so, when the TVD is experimentally tested, there are significant discrepancies in the frequency response function (FRF), due to simplifications and model assumptions. Here, we implement the FRF as a deep neural network using a direct graph. The model elements, such as storage and loss moduli, stiffness and damping coefficients are nodes of this graph. Then, we add data-driven nodes (implemented as multilayer perceptrons) to correct the outputs of the stiffness and damping coefficients. This way, the gap between predicted and observed FRF can be closed. With this framework, we can build hybrid models that merge the original computer model (or at least, a reduced-order representation of it) with the neural network through a graph. This allows us to estimate the model-form uncertainty even for hidden nodes of the graph. In the TVD application, we studied the performance of our framework both in interpolation (when the model predicts the FRF between observations) and extrapolation (when the model predicts the FRF outside the observation range). The results demonstrate the ability to perform simultaneous estimation of discrepancy at reasonable computational cost.

Restricted internal oxygen isotope exchange in calcite veins: Constraints from fluid inclusion and clumped isotope-derived temperatures

The distribution of oxygen isotopes between calcite and fluid inclusions has demonstrated utility for reconstructing near-surface calcite precipitation temperatures. For calcite that formed at depth, however, the resilience of this paleothermometer to diagenetic oxygen isotope alteration is poorly constrained. Clumped isotopes also document calcite precipitation temperatures and are similarly vulnerable to diagenetic alteration. Post-entrapment isotope exchange between calcite and fluid-inclusions could alter the calcite-fluid oxygen isotope distribution (αc-fi), as well as the clumped isotope composition (Δ47) of calcite, and therefore these two seemingly independent paleo-thermometers are potentially linked via the same alteration process. Using closed-system batch fractionation equations, we have modelled various scenarios of oxygen isotope exchange between water and host-rock during burial, as well as internal oxygen isotope exchange between calcite and fluid inclusions during exhumation. Assuming both paleo-thermometers record concordant temperatures at the time of vein formation, our models predict that if a fraction of calcite is available for isotopic interaction with fluid inclusions, the fluid inclusion and clumped isotope-derived paleothermometers yield discrepant temperature estimates after exhumation. We show that the fluid inclusion thermometer is more sensitive to isotopic alteration than the clumped isotope thermometer and that the mass balance of oxygen between calcite and fluid inclusions determines the sensitivity of both paleothermometers as well as the vulnerability of fluid inclusions (δ18Ofi) to diagenetic overprinting. We applied coupled clumped isotope and fluid inclusion measurements on calcite veins from the External Albanides (Albania), which were formed at depth and subsequently exhumed, in order to compare natural samples to our isotope exchange model. These veins show strongly discrepant calcite-water equilibrium temperatures and clumped isotope temperatures, suggesting the fraction of calcite available for isotope exchange with internal fluids may indeed be a key parameter of diagenetic alteration during exhumation. Even though the clumped isotope temperatures of our samples appear to be insensitive towards internal oxygen isotope exchange, our model predicts that at low burial temperatures, the carbonate clumped isotope thermometer may be susceptible to alteration by diagenetic isotope exchange with fluid inclusions under certain conditions.

Estimating model inadequacy in ordinary differential equations with physics-informed neural networks

A number of physical systems can be described by ordinary differential equations. When physics is well understood, the time dependent responses are easily obtained numerically. The particular numerical method used for integration depends on the application. Unfortunately, when physics is not fully understood, the discrepancies between predictions and observed responses can be large and unacceptable. In this paper, we propose an approach that uses observed data to estimate the missing physics in the original model (i.e., model-form uncertainty). In our approach, we first design recurrent neural networks to perform numerical integration of the ordinary differential equations. Then, we implement the recurrent neural network as a directed graph. This way, the nodes in the graph represent the physics-informed kernels found in the ordinary differential equations. We quantify the missing physics by carefully introducing data-driven in the directed graph. This allows us to estimate the missing physics (discrepancy term) even for hidden nodes of the graph. We studied the performance of our proposed approach with the aid of three case studies (fatigue crack growth, corrosion-fatigue crack growth, and bearing fatigue) and state-of-the-art machine learning software packages. Our results demonstrate the ability to perform estimation of discrepancy, reducing gap between predictions and observations, at reasonable computational cost.

Optical performance monitoring using digital coherent receivers and convolutional neural networks.

We experimentally demonstrate accurate modulation format identification, optical signal to noise ratio (OSNR) estimation, and bit error ratio (BER) estimation of optical signals for wavelength division multiplexed optical communication systems using convolutional neural networks (CNN). We assess the benefits and challenges of extracting information at two distinct points within the demodulation process: immediately after timing recovery and immediately prior to symbol unmapping. For the former, we use 3D Stokes-space based signal representations. For the latter, we use conventional I-Q constellation images created using demodulated symbols. We demonstrate these methods on simulated and experimental dual-polarized waveforms for 32-GBaud QPSK, 8QAM, 16QAM, and 32QAM. Our results show that CNN extracts distinct and learnable features at both the early stage of demodulation where the information can be used to optimize subsequent stages and near the end of demodulation where the constellation images are readily available. Modulation format identification is demonstrated with >99.8% accuracy, OSNR estimation with <0.5 dB average discrepancy and BER estimation with percentage error of <25%.