Accurate Performance Estimation Research Articles

LAE-GAN-Based Face Image Restoration for Low-Light Age Estimation

Age estimation is applicable in various fields, and among them, research on age estimation using human facial images, which are the easiest to acquire, is being actively conducted. Since the emergence of deep learning, studies on age estimation using various types of convolutional neural networks (CNN) have been conducted, and they have resulted in good performances, as clear images with high illumination were typically used in these studies. However, human facial images are typically captured in low-light environments. Age information can be lost in facial images captured in low-illumination environments, where noise and blur generated by the camera in the captured image reduce the age estimation performance. No study has yet been conducted on age estimation using facial images captured under low light. In order to overcome this problem, this study proposes a new generative adversarial network for low-light age estimation (LAE-GAN), which compensates for the brightness of human facial images captured in low-light environments, and a CNN-based age estimation method in which compensated images are input. When the experiment was conducted using the MORPH, AFAD, and FG-NET databases—which are open databases—the proposed method exhibited more accurate age estimation performance and brightness compensation in low-light images compared to state-of-the-art methods.

Real-Time QoT Estimation Through SDN Control Plane Monitoring Evaluated in Mininet-Optical

Quality of Transmission (QoT) metrics are used to predict and evaluate optical transmission performance, and have received much attention due to recent interest in real-time software defined networking (SDN) control and the challenges posed by open, disaggregated optical systems. In this letter, we address the QoT-estimation (QoT-E) issue, by focusing on real-time operations, where a dynamic control plane improves path performance estimation accuracy by interrogating optical performance monitoring (OPM) devices in the network. We also recognize the difficulty in testing such control plane procedures over large scale systems, which is a necessary step to validate control plane scalability before it is implemented in real systems. Accordingly, we have developed an optical network simulation system that incorporates physical transmission modeling into the Mininet packet-network emulator to enable the development and evaluation of optical control plane processes that depend on optical transmission performance. In this letter, we use this new tool, Mininet-Optical, to evaluate a SDN QoT-E procedure based on the deployment of emulated OPM nodes at periodic locations in an optical transmission system, which are interrogated by a SDN controller in real-time. Through this new capability, we observe behavior that manifests across a large scale system of 70 in line amplifiers and 15 ROADM nodes and are able show improvements in QoT accuracy up to 3 dB.

Intellectual capital and firm efficiency of US multinational software firms

PurposeThe main purpose of this study is to empirically investigate the impact of intellectual capital efficiency on US multinational software companies' performance from 2012 to 2016 by applying data envelopment analysis (DEA).Design/methodology/approachIt adopts a new slacks-based measure (SBM) to obtain a more accurate performance estimation and rank between companies. Regression analysis is used to test the overall IC and each of its elements (Human Capital, Innovation Capital, Process Capital and Customer Capital).FindingsThe univariate result shows that multinational companies are more efficient than non-multinational companies. However, the regression result shows that multinationality can hardly explain the firm efficiency of software firms. Another interesting finding is that intellectual capital has a positive and significant impact on software firm performance in the US human capital influences firm efficiency directly. However, when human capital is combined with the other elements of IC, the contribution of human capital becomes less significant. This is because people may think that innovation capital, process capital and customer capital can replace human capital, but it is not. In short, human capital may affect firm efficiency through other elements of IC (innovation capital, process capital and customer capital) as it is the base of other elements.Research limitations/implicationsThe results show that multinational companies have higher efficiency scores than non-multinational companies. In addition, Intellectual capital has a positive and significant impact on software firm performance in the US human capital influences firm efficiency directly. However, when human capital is combined with the other elements of IC, the contribution of human capital becomes less significant. This is because people may think that innovation capital, process capital and customer capital can replace human capital, but it is not. In short, human capital may affect firm efficiency through other elements of IC (innovation capital, process capital and customer capital) as it is the base of other elements.Practical implicationsOverall, the study highlights the needs of having intellectual capital and its elements (Human Capital, Innovation Capital, Process Capital and Customer Capital) to increase firm efficiency.Originality/valueFirst, the authors use a more comprehensive elements of IC, which are human capital, innovation capital, process capital and customer capital for a better IC measurement. Second, this study makes the first attempt using the DSBM model via DEA to examine the operating efficiency of US multinational software firms.

Optimal estimation of snow and ice surface parameters from imaging spectroscopy measurements

Snow and ice melt processes are a key in Earth's energy-balance and hydrological cycle. Their quantification facilitates predictions of meltwater runoff as well as distribution and availability of fresh water. They control the balance of the Earth's ice sheets and are acutely sensitive to climate change. These processes decrease the surface reflectance with unique spectral patterns due to the accumulation of liquid water and light absorbing particles (LAP), that require imaging spectroscopy to map and measure. Here we present a new method to retrieve snow grain size, liquid water fraction, and LAP mass mixing ratio from airborne and spaceborne imaging spectroscopy acquisitions. This methodology is based on a simultaneous retrieval of atmospheric and surface parameters using optimal estimation (OE), a retrieval technique which leverages prior knowledge and measurement noise in an inversion that also produces uncertainty estimates. We exploit statistical relationships between surface reflectance spectra and snow and ice properties to estimate their most probable quantities given the reflectance. To test this new algorithm we conducted a sensitivity analysis based on simulated top-of-atmosphere radiance spectra using the upcoming EnMAP orbital imaging spectroscopy mission, demonstrating an accurate estimation performance of snow and ice surface properties. A validation experiment using in-situ measurements of glacier algae mass mixing ratio and surface reflectance from the Greenland Ice Sheet gave uncertainties of ±16.4 μg/gice and less than 3%, respectively. Finally, we evaluated the retrieval capacity for all snow and ice properties with an AVIRIS-NG acquisition from the Greenland Ice Sheet demonstrating this approach's potential and suitability for upcoming orbital imaging spectroscopy missions.

Field Computation in Media Exhibiting Hysteresis Using Hopfield Neural Networks

Assessment of local magnetization in objects exhibiting hysteresis is crucial to the accurate design and performance estimation for a wide range of electromagnetic devices. In the past, different efforts have been carried out to incorporate hysteresis models in field computation approaches. While those models varied in methodologies, they shared a common goal of offering an accurate and computationally efficient field computation tool. Recently, it was demonstrated that a two-dimensional vector hysteresis operator may be realized using a tri-node Hopfield neural network. The purpose of this paper is to offer a novel two-dimensional field computation approach in media exhibiting hysteresis that utilizes the aforementioned hysteresis operator. The approach is based on incorporating domain-to-domain interactions in the overall network energy formulation while employing typical Hopfield neural network energy minimization algorithms. Details of the proposed model, numerical simulations and comparisons with finite-element calculations are given in the paper.

On-Orbit Pulse Phase Estimation Based on CE-Adam Algorithm

Pulse phase is the basic measurements of X-ray pulsar-based navigation, and thus how to estimate a pulse phase for an orbiting spacecraft is important. The current methods for on-orbit pulse phase estimation could provide an accurate estimation performance enhancing with the photon amount, but its central processing unit (CPU) time cost also increases sharply with the increase of photon amount. In this paper, an on-orbit pulse phase estimation method based on the cross-entropy adaptive moment estimation (CE-Adam) algorithm is proposed to reduce the CPU time cost while retaining decent estimation accuracy. This method combines the CE and Adam algorithms, and is able to obtain a global optimum with low CPU time cost. The performance of the proposed algorithm is verified by simulation data and real data from the Neutron Star Internal Composition Detector (NICER). The results show that the proposed algorithm could greatly reduce the CPU time cost, which is about 1.5% of the CE algorithm, and retain similar estimation accuracy of pulse phase with CE algorithm.

An Online Flux Estimation for Dual Three-Phase SPMSM Drives Using Position-Offset Injection

The rotor flux linkage of permanent-magnet synchronous motor (PMSM) is of great importance for monitoring and high-performance control. However, most of the existing online estimation methods need prior knowledge of other PMSM parameters such as stator resistance and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dq</i> axis inductance, which are difficult to realize in the rank-deficient models of electrical machines. This article proposes a simple online estimation method of rotor flux linkage for the dual three-phase surface-mounted PMSM (SPMSM) drives with the assistance of injected position offset. The independent estimation model has been developed for the rotor flux linkage by appropriate setting of positon offset between two sets of three-phase windings of dual three-phase SPMSM drives. The proposed method not only has little influence on the output torque and high signal-to-noise ratio for observation, but also offers the superior performance in estimation accuracy irrespective of the influence of VSI nonlinearity. In addition, the proposed position-offset-based method is suitable for the full-speed region including the field-weakening operation. Finally, comprehensive experimental results are presented to verify validity of the proposed method under different working conditions.

Employing load and irradiance profiles for the allocation of PV arrays with inverter reactive power and battery storage in distribution networks–A fast comprehensive QSTS technique

The paper presents a novel computationally efficient Quasi-Static-Time-Series (QSTS) approach for the sizing and sitting of photovoltaic arrays in a distribution network, which uses historical hourly demand and irradiance data. Additionally, the proposed approach combines a battery energy storage system (BESS) with the PV array for smoothing the array output and integrates inverter reactive power control for voltage support. An optimized search method based on the combined Non-dominated Sorting Genetic Algorithm-II and Fuzzy Decision-Making Tool (NSGAII/FDMT) is employed for the sizing and sitting problem along with three objectives: (a) minimization of total cost; (b) minimization of distribution network power losses; and (c) maximization of system security. The distribution network-PV array-BESS system along with its control functions is solved in hourly steps for a yearly 8760-h horizon obtaining an accurate estimate of the system performance. In order to decrease execution time, a robust formulation of the direct power flow method is developed in the paper yielding a fast 8760-solution algorithm. The performance of the proposed method is tested through different case studies on a modified IEEE 123-bus and 33-bus radial distribution systems. The results demonstrate that detailed system analysis using historical data produces more efficient options than conventional methods based on system analysis at peak load only.

Utilizing four-node tetrahedra-shaped Hopfield neural network configurations in the local magnetization assessment of 3d objects exhibiting hysteresis

There is no doubt that the accurate assessment of local magnetization in three-dimensional objects exhibiting hysteresis is crucial to accurate performance estimation of a variety of electromagnetic devices. Recently, it has been demonstrated that a Stoner-Wohlfarth-like elementary hysteresis operator may be constructed using two-node Hopfield neural network (HNN) having internal positive feedback. Based upon the previously mentioned approach, this paper presents a methodology using which local magnetization in 3D objects exhibiting hysteresis may be assessed. The approach utilizes a four-node tetrahedron-shaped HNN with activation functions constructed using a weighted superposition of a step and sigmoidal functions in accordance with the M − H curve of the material under consideration. In this approach, the internal feedback factors between the different nodes for any tetrahedron are dependent on its geometrical configuration. Hence, shape configuration effects on the magnetization patterns of any three-dimensional object approximated by an ensemble of tetrahedra are implicitly taken into consideration. To demonstrate the applicability of the proposed approach, computations were carried out for different three-dimensional magnetized bodies having different M − H curves. Theoretical and computational details of the approach are given in the paper.

Innovation Feedback Kalman Filter

The conventional Kalman filter demands complete knowledge of the actual filter model, which is usually inaccessible in practical problems. When dealing with the filtering problem with uncertainty, any improper prediction of the uncertainty may degrade the filtering performance and even lead to divergence phenomenon. In this paper, the innovation feedback Kalman filter, which introduces the innovation feedback controller to the Kalman filter equations, is newly proposed on the basis of automatic control theory to address the filtering problem with uncertainty and is different from other filtering methods. By studying the estimate error equations, the estimate bias is extracted and its propagation mechanism is formulated. The estimate bias propagation equations reveal that eliminating the estimate bias essentially equivalents to an output regulation problem with uncertain exosystem. And a concise yet effective innovation feedback controller is subsequently given as an example. The proposed method is applied to one-dimension target tracking scenario and its high estimation accuracy performance and good stability are simulated through a comparative analysis with the ideal filtering results.

Compact modeling of metal-oxide TFTs based on artificial neural network and improved particle swarm optimization

The application of artificial neural network (ANN) can give a very accurate and fast model for semiconductor devices used in circuit simulations. In this paper, we have applied multi-layer perceptron (MLP) neural network based on limited memory Broyden–Fletcher–Goldfarb–Shanno (L-BFGS) method to model the flexible metal-oxide thin-film transistors (TFTs). An improved particle swarm optimization (PSO) is employed to find suitable initial parameters for the ANN model, which consists of a centroid opposition-based learning algorithm and a mutation strategy based on Euclidean distance to enhance the searching ability further. This hybrid modeling routine can improve the accuracy of predictions of both the I–V and small signal parameters ( $$g_d$$ , $$g_m$$ , etc.) characteristics, which are in good agreement with experimental data and fully demonstrate the validity of the proposed model. Furthermore, the model is implemented into a simulator with Verilog-A. The circuit-level tests of TFT show that the ANN compact model with PSO enables accurate performance estimation of metal-oxide TFT circuits.

A practical method for determining automated EEG interpretation software performance on continuous Video-EEG monitoring data

Despite evidence to suggest that as many as 30% of hospitalized patients with altered mental status may be having subclinical seizures for which treatment may improve patient outcomes, the adoption of continuous video electroencephalography (EEG) monitoring (cvEEG) into non-academic centers has remained slow. With the continued progress in EEG networking technology and telemedicine, the traditional barriers to the use of cvEEG in community hospitals may soon be removed. This will lead to significant increases in the volume of cvEEG data that needs to be reviewed. The application of computer analytics to automate the process of cvEEG review has been ongoing for many years. One solution has been the use of graphic trends to allow more rapid review and assessment of the content of the cvEEG data over long periods of time. However, high false positive rates, the lack of specificity, and the requirement for frequent review of the raw cvEEG data to confirm the results has limited the use of this modality as a bedside monitoring tool. Thus, the development of truly automated cvEEG readers will be needed. Here we address the challenges of assessing the performance of such automated readers using the traditional core measures of performance: sensitivity, specificity, false positive/negative rates and diagnostic accuracy, and introduce an alternative approach to measuring and assessing performance that focuses on agreement. Our work demonstrates that traditional approaches to objectively assess agreement and performance of automated readers are difficult to apply and may not provide accurate estimates of the real-world performance of computer analytic interpretations of cvEEG data. We introduce our “Pure Gold” approach to the assessment of automated cvEEG reading tools. This novel approach uses compiled EEG segments representing a single target pattern to measure true error rates for the automated reader, visual graphical representation of automated reader outputs for comparison to human reviews of the same data, and agreement comparisons with correction for human variation. We suggest our approach represents a better method for performance assessment than traditional metrics of sensitivity and specificity.

GPGPU Performance Estimation With Core and Memory Frequency Scaling

Graphics Processing Units (GPUs) support dynamic voltage and frequency scaling (DVFS) in order to balance computational performance and energy consumption. However, there still lacks simple and accurate performance estimation of a given GPU kernel under different frequency settings on real hardware, which is important to decide best frequency configuration for energy saving. This paper reveals a fine-grained model to estimate the execution time of GPU kernels with both core and memory frequency scaling. Over a 2.5x range of both core and memory frequencies among 12 GPU kernels, our model achieves accurate results (within 3.5\%) on real hardware. Compared with the cycle-level simulators, our model only needs some simple micro-benchmark to extract a set of hardware parameters and performance counters of the kernels to produce this high accuracy.

Support vector machine-based classification of schizophrenia patients and healthy controls using structural magnetic resonance imaging from two independent sites.

Structural brain alterations have been repeatedly reported in schizophrenia; however, the pathophysiology of its alterations remains unclear. Multivariate pattern recognition analysis such as support vector machines can classify patients and healthy controls by detecting subtle and spatially distributed patterns of structural alterations. We aimed to use a support vector machine to distinguish patients with schizophrenia from control participants on the basis of structural magnetic resonance imaging data and delineate the patterns of structural alterations that significantly contributed to the classification performance. We used independent datasets from different sites with different magnetic resonance imaging scanners, protocols and clinical characteristics of the patient group to achieve a more accurate estimate of the classification performance of support vector machines. We developed a support vector machine classifier using the dataset from one site (101 participants) and evaluated the performance of the trained support vector machine using a dataset from the other site (97 participants) and vice versa. We assessed the performance of the trained support vector machines in each support vector machine classifier. Both support vector machine classifiers attained a classification accuracy of >70% with two independent datasets indicating a consistently high performance of support vector machines even when used to classify data from different sites, scanners and different acquisition protocols. The regions contributing to the classification accuracy included the bilateral medial frontal cortex, superior temporal cortex, insula, occipital cortex, cerebellum, and thalamus, which have been reported to be related to the pathogenesis of schizophrenia. These results indicated that the support vector machine could detect subtle structural brain alterations and might aid our understanding of the pathophysiology of these changes in schizophrenia, which could be one of the diagnostic findings of schizophrenia.

Water vapor effects on CO2 separation of amine‐containing facilitated transport membranes (AFTMs) module: mathematical modeling using tanks‐in‐series approach

AbstractAmine‐containing facilitated transport membranes (AFTMs) are emerging membranes and have great potential for CO2 capture. AFTMs are particularly suitable for humid flue gases, as water vapor enhances the facilitated transportation of CO2. However, there is a lack of mathematical models available to evaluate the impact of water vapor on CO2 separation performance for large‐scale membrane modules. Therefore, developing a mathematical model considering the water vapor effects is of great importance to guide the large‐scale applications of AFTMs for CO2 capture.In this study, a mathematical model using tanks‐in‐series approach for CO2 capture through AFTMs is proposed. In this model, the tanks‐in‐series approach can reflect the variable (water vapor‐dependent) gas permeance along the membrane module. Modeling results showed the necessity of considering the effect of water vapor content decline along the module. In addition, both increasing temperature and pressure primarily lead to a faster decline of relative humidity (RH) along the membrane module, resulting in an increase of CO2 recovery but a decrease of CO2 product purity. For a gas stream with relatively low water vapor content (&lt;2% in this study), different H2O permeance has a minor effect on separation performance. Moreover, permeate gas with higher water vapor content as well as higher RH is considered more suitable for further treatment of second stage membrane separation. In conclusion, the proposed model provides insights for comprehensive parameter analysis of membrane CO2 capture simulation and could support the development of large‐scale AFTMs separation applications, particularly for systems design and more accurate performance estimation. © 2020 Society of Chemical Industry and John Wiley &amp; Sons, Ltd.

A novel academic performance estimation model using two stage feature selection

&lt;span lang="EN-GB"&gt;Educational data mining has gained tremendous interest from researchers across the globe. Using data mining techniques in the field of education several significant findings have been made. Accurate academic performance estimation is a challenging task. In this study we have developed a novel model to estimate the academic performance of students. Techniques like conversion of categorical attributes into dummy variables, classification, two staged feature selection and an improved differential evolutionary algorithm were used. Our proposed model outperformed existing models of students’ academic performance determination and gave a new direction to it. The proposed model can help not only to reduce the number of academic failures but also help to comprehend the factors contributing to a student’s academic performance (poor, average or outstanding).Computer&lt;/span&gt;

Augmented inverse probability weighted fractional imputation in quantile regression.

By employing all the observed information and the optimal augmentation term, we propose an augmented inverse probability weighted fractional imputation method (AFI) to handle covariates missing at random in quantile regression. Compared with the existing completely case analysis, inverse probability weighting, multiple imputation and fractional imputation based on quantile regression model with missing covarites, we carry out simulation study to investigate its performance in estimation accuracy and efficiency, computational efficiency and estimation robustness. We also talk about the influence of imputation replicates in our AFI. Finally, we apply our methodology to part of the National Health and Nutrition Examination Survey data.

Ranking and selection for terminating simulation under sequential sampling

This research develops an efficient ranking and selection procedure to select the best design for terminating simulation under sequential sampling. This approach enables us to obtain an accurate estimate of the mean performance at a particular point using regression in the case of a terminating simulation. The sequential sampling constraint is imposed to fully utilize the information along the simulation replication. The asymptotically optimal simulation budget allocation among all designs is derived concurrently with the optimal simulation run length and optimal number of simulation groups for each design. To implement the simulation budget allocation rule with a fixed finite simulation budget, a heuristic sequential simulation procedure is suggested with the objective of maximizing the probability of correct selection. Numerical experiments confirm the efficiency of the procedure relative to extant approaches.

A comparative study for fatigue characterization of asphalt binder using the linear amplitude sweep test

Currently the linear amplitude sweep (LAS) test is one of the main fatigue protocols to estimate the fatigue resistance of asphalt binder. In past 10-years, several data interpretation approaches have been developed for the LAS-based binder fatigue modeling. The objective of this paper is to compare the currently available analytical approaches for the data interpretation of LAS test and further investigate the potential links between the different LAS-based fatigue evaluations of asphalt binders. Totally 20 types of unmodified and modified asphalt binders are selected in this study, which actually cover 35 sample conditions under different aging and temperature levels. Experimental and analysis results indicate that the full fatigue performance of asphalt binder can be characterized from with the S-VECD & GR approach. The failure property in LAS test is also well correlated to the binder fatigue life simulation and thus can be utilized to more efficiently distinguish the binder fatigue resistance. Furthermore, the performance predictive accuracy of AASHTO TP101 specification method is strongly dependent on the selected failure definition. Taking the artificial failure definition of 35% reduction in |G*|·sin δ largely underestimates the binder fatigue life when compared to the S-VECD & GR approach. However, when a material-dependent failure definition of peak stress is alternatively employed in specification method, the correlation of the predictive binder fatigue life between the two analytical approaches is significantly improved and a linear relationship is additionally found between the S-VECD & GR approach and specification method. Therefore, it is recommended that when only the standard LAS test with one loading rate is available in practice, the peak stress failure definition should be used in AASHTO TP101 specification method to provide relatively accurate fatigue performance estimation of asphalt binder.

Direct data domain space-time adaptive processing method based on joint sparse representation with single snapshot for MIMO radar

To improve the performance of exiting sparse recovery-based space-time adaptive processing (STAP) methods in nonstationary and heterogeneous environments, a STAP method, called the joint sparse representation (JSR)-based direct data domain STAP method using MIMO subarraying radar, is proposed. First, by exploiting the property of waveforms orthogonal diversity of MIMO subarraying radar, the received single snapshot signal of the range cell under test is expressed as multiple snapshot signals. Second, the JSR model of clutter plus target spatiotemporal spectrum estimation is established and the joint sparsity is proved. The spatiotemporal spectrum of clutter and target is then estimated by the typical JSR algorithm. At last, the clutter covariance clutter matrix and the corresponding weight vector of the STAP processor are calculated based on the clutter space-time distribution extracted by the rough priori knowledge of the target. By utilizing the data of the range cell under test only, the proposed method could achieve accurate spatiotemporal spectrum estimation and great performance of clutter suppression. Experimental simulations demonstrate the validity of the proposed method.