Models In Terms Of Robustness Research Articles

Robust Statistical Modeling of Heterogeneity for Repairable Systems Using Multivariate Gaussian Convolution Processes

A main challenge in reliability analysis of repairable systems is to model the heterogeneity in their failure behavior, which can be reflected by the corresponding recurrent failure-time data. To capture the system heterogeneity for data analysis, a system-specific random effect is typically introduced in most existing statistical models. In practice, the random effect of repairable systems tends to be time varying; for example, each repair action could change system's physical properties. Prior studies, however, generally do not take account of this time-varying nature and few of them circumvent the risk of model misspecification on the parametric distribution of frailty. In this article, we propose a semiparametric model that uses multivariate Gaussian convolution processes (MGCPs) to meet the above challenges. First, we use the trend RP to model the baseline intensity function of each repairable system. Based on the baseline intensity function, we then introduce MGCPs to simultaneously factor in heterogeneity and infer commonalities across multiple systems. A Bayesian framework is used for parameter estimation and time-to-failure prediction. Simulation studies show the advantages of our model in terms of robustness and estimation accuracy. A group of oil and gas well systems are used to illustrate the application of the proposed model.

Integrated optimization of assortment, inventory and pricing considering omnichannel retailer’s risk aversion and customer’s time preference

This paper mainly discusses the joint optimization of product assortment, inventory and pricing for an omnichannel retailer under uncertain demand, seeking to maximize the retailer’s Worst-case Conditional Value-at-Risk (WCVaR) based profit and customer’s expected utility. The risk aversion of decision-maker and the time preference of customers are depicted by the WCVaR and quasi-hyperbolic discounting function, respectively. A bi-objective stochastic optimization model is developed. To alleviate the computational burden, a linearization method is applied to convert the problem into a mixed integer linear programming that proves to be solvable within reasonable CPU times using the augmented ε-constraint method. Numerical studies are conducted to investigate the applicability of the proposed model and the efficiency of the solution approach. Our experimental results show that offering customers with high expected utility typically requires low prices and small assortments. We also make other counterintuitive observation that the price of products sometimes increases when the expected utility increases. Furthermore, WCVaR has superiority compared with the risk-neutral model in terms of robustness and stability. The time preference of customers has non-negligible impacts on decision-making results. In addition, the further sensitivity analyses investigate the impacts of various key parameters on the performance of omnichannel retailing.

Analysis of the attack effect of adversarial attacks on machine learning models

The use of neural networks has produced outstanding results in a variety of domains, including computer vision and text mining. Numerous investigations in recent years have shown that using adversarial attacks technology to perturb the input samples weakly can mislead most mainstream neural network models, for example Fully Connected Neural Networks (FCNN) and Convolutional Neural Networks (CNN), to make wrong judgment results. Adversarial attacks can help researchers discover the potential defects of neural network models in terms of robustness and security so that people can comprehend the neural network models' learning process better and solve the neural network models' interpretability. However, suppose an adversarial attack is performed on a non-deep learning model. In that case, the results are very different from the deep learning model. This paper first briefly outlines the existing adversarial example technology; then selects the CIFAR10 dataset as the test data and LeNet, ResNet18, and VGG16 as the test model according to the technical principle; then uses the Fast Gradient Sign Attack (FGSM) method to conduct attack experiments with the CNNs and traditional machine learning algorithms like K-Nearest Neighbors (KNN) and Support Vector Machine (SVM); then analyze the experimental results and find that the adversarial example technology is specific to the deep learning model, but it cannot be completely denied that adversarial examples have no attack effect on traditional machine learning models.

MemeTector: enforcing deep focus for meme detection

Image memes and specifically their widely known variation image macros are a special new media type that combines text with images and are used in social media to playfully or subtly express humor, irony, sarcasm and even hate. It is important to accurately retrieve image memes from social media to better capture the cultural and social aspects of online phenomena and detect potential issues (hate-speech, disinformation). Essentially, the background image of an image macro is a regular image easily recognized as such by humans but cumbersome for the machine to do so due to feature map similarity with the complete image macro. Hence, accumulating suitable feature maps in such cases can lead to deep understanding of the notion of image memes. To this end, we propose a methodology, called visual part utilization, that utilizes the visual part of image memes as instances of the regular image class and the initial image memes as instances of the image meme class to force the model to concentrate on the critical parts that characterize an image meme. Additionally, we employ a trainable attention mechanism on top of a standard ViT architecture to enhance the model’s ability to focus on these critical parts and make the predictions interpretable. Several training and test scenarios involving web-scraped regular images of controlled text presence are considered for evaluating the model in terms of robustness and accuracy. The findings indicate that light visual part utilization combined with sufficient text presence during training provides the best and most robust model, surpassing state of the art. Source code and dataset are available at https://github.com/mever-team/memetector.

Semantic Representation Fusion-Based Network for Robust Land Cover Classification in Foggy Conditions

A precise and robust classification of land cover is crucial for land use estimation. A robust model that can provide rich semantic information is imperative for the challenging task of land cover classification in foggy conditions. We propose Semantic Representation Enhancement ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">SRE</i> ) and Semantic Representation Aggregation ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">SRA</i> ) modules for the fusion of semantic representation. The Dense Depthwise Separable Atrous Spatial Pyramid Pooling ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DDS-ASPP</i> ) module in SRE possesses a large receptive field, which covers an extensive scale range. Enhanced asymmetric convolution module ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">EACM</i> ) in SRE focus on features of various directions. DDS-ASPP and EACM generate the class-based and pixel-based representation respectively. By means of SRA and dual representations, we model the relationship between global context and coarse class regions to capture long-range correlation. Moreover, evaluated on Potsdam, Vaihingen and custom real-world datasets under fog, we demonstrate that our work is competitive with state-of-the-art models in terms of robustness. Code will be available at https://github.com/bowenroom/Robust-land-cover-classification.

Fairness Measures of Machine Learning Models in Judicial Penalty Prediction

&lt;p&gt;Machine learning (ML) has been widely adopted in many software applications across domains. However, accompanying the outstanding performance, the behaviors of the ML models, which are essentially a kind of black-box software, could be unfair and hard to understand in many cases. In our human-centered society, an unfair decision could potentially damage human value, even causing severe social consequences, especially in decision-critical scenarios such as legal judgment. Although some existing works investigated the ML models in terms of robustness, accuracy, security, privacy, quality, etc., the study on the fairness of ML is still in the early stage. In this paper, we first proposed a set of fairness metrics for ML models from different perspectives. Based on this, we performed a comparative study on the fairness of existing widely used classic ML and deep learning models in the domain of real-world judicial judgments. The experiment results reveal that the current state-of-the-art ML models could still raise concerns for unfair decision-making. The ML models with high accuracy and fairness are urgently demanding.&lt;/p&gt; &lt;p&gt;&amp;nbsp;&lt;/p&gt;

A Variational Level Set Image Segmentation Method via Fractional Differentiation

To solve the issues with conventional level set segmentation algorithms, which are sensitive to the initial contours and less noise-resistant, a segmentation model based on the coupling of texture information and structural information is developed. In this model, a rotation invariant mask produced by fractional-order differentiation is used to first describe the image’s global information. Then, the power function of the energy generalization function is solved by applying factorization theory, and for each pixel of the image, not only its information but also its surrounding pixel information is taken into account and integrated into the energy generalization function via weight scaling. At the same time, the L2 norm of the fractional-order image and the difference from the fitted image are used to generate the energy generalization function of the model. The final results of this study demonstrate that the proposed model achieved a better segmentation performance than the current active contour models in terms of robustness to Gaussian noise and pretzel noise, as well as the segmentation accuracy and algorithm running time. These results were obtained in synthetic images, real images, and natural images.

Predictive modeling of compressive strength of sustainable rice husk ash concrete: Ensemble learner optimization and comparison

One of the largest sources of greenhouse gas (GHG) emissions is the construction concrete industry which has alone 50% of the world's emissions. One possible remedy to mitigate the effect of environmental issues is the use of waste and recycled material in concrete. Today, immense agricultural waste is being used as a substitute for cement in the production of sustainable concrete. Therefore, this study is aimed to predict and develop an empirical formula of the compressive strength of rice husk ash (RHA) concrete using machine learning algorithms. Methods employed in this study includes gene expression programming (GEP) and Random Forest Regression (RFR). A reliable database of 192 data points was employed for developing the models. Most influential variables including age, cement, rice husk ash, water, super plasticizer, and aggregate were employed as input parameters in the development of RHA-based concrete models. Evaluation of models was performed using different statistical parameters. These statistical measures include mean absolute error (MAE), coefficient of determination (R2), performance index (ρ), root man square error (RMSE), relative squared error (RSE) and relative root mean square (RRMSE). The GEP model outperforms the RFR ensemble model in terms of robustness, with a greater correlation of R2 = 0.96 compared to RFR's R2 = 0.91. Ensemble modeling showed an enhancement of 1.62 percent for RFR compressive strength model when compared with individual RFR compressive strength model as illustrated by statistical parameters. Moreover, GEP model shows an enhancement of 37.33 percent in average error with an average error 2.35 MPa as compared to RFR model with average error of about 3.75 MPa. Cross validation was used as external check to avoid overfitting issues of the models and confirm the generalized model output. Parametric analysis was performed to determine the impact of the input parameters on the output. Cement and age were shown to have a substantial impact on the compressive strength of RHA concrete using sensitivity analysis.

Adaptive Under-Sampling Deep Neural Network for Rapid and Reliable Image Recovery in Confocal Laser Scanning Microscope Measurements

Confocal laser scanning microscopy (CLSM) is a non-destructive optical measurement system of high precision, applicable to the construction of three-dimensional (3-D) topographies of biological cells and engineered materials at the micro- and sub-micro scales. Compressive sensing (CS) has recently been applied in microscope systems to reduce the amount of sampled data required for the reconstruction of images; however, the iterative nature of the CS recovery algorithm imposes high computational complexity. This article presents an end-to-end non-iterative deep residual convolutional neural network (CNN) applicable to CLSM systems for CS-based reconstruction. In experiments and numerical simulations, the proposed scheme outperformed the existing CS recovery algorithms in terms of reconstructed image quality and computation time. The proposed algorithm also enabled the reconstruction of images using samples obtained in different regions of an image at various sampling rates to overcome nonuniform information density. The reconstruction performance of the model in terms of robustness and efficiency was validated using real-world CLSM data obtained via random scanning patterns.

Investigation of optimal vegetation indices for retrieval of leaf chlorophyll and leaf area index using enhanced learning algorithms

With the availability of high-resolution data due to sensor technology advancement, it is now easier for researchers and scientists to detect or view the spectral variability of different crops. For this study, Leaf chlorophyll content (LCC) and Leaf area index (LAI) of the crops Maize (Zea mays), Mustard (Brassica), and pink Lentils (Lens esculenta) under different irrigation and fertilizer treatments have been analyzed. In total, rigorous assessment of 25-hyperspectral vegetation indices (VIs) at both leaf and canopy level for chlorophyll content, whereas 7- hyperspectral VIs for LAI at canopy level were computed to investigate the robustness of these VIs for LCC and LAI assessment. Variable importance in projection (VIP) using Partial Least Square regression (PLSR) and coefficient of determination (R2) were computed for all the VIs to extract the most sensitive information for the retrieval of LCC and LAI. As a result, the VIs using the red-edge reflectance bands at 705 and 750 nm were found highly responsive to LAI compared to other wavebands. In contrast, the VIs indices made of green (550 nm), red (670, 690, and 700 nm), and red-edge (705, 750 nm) bands were found highly sensitive to the temporal LCC values of lentils and maize crop beds. In addition, the temporal LCC values of Mustard crop beds’ were found sensitive to the VIs made of green (550 nm), red (670, 690, and 700 nm), and NIR (800 nm) wavebands. The three VIs having high VIP and R2 values were selected as optimum sets of input to build support vector regression models using radial (SVR-Rad), linear (SVR-Li), polynomial (SVR-Poly), Random Forrest Regression (RFR), Partial least square regression (PLSR), and Hybrid neural fuzzy inference system (HyFIS). The analysis showed that the SVR-Rad model outperformed the SVR-Li, SVR-Poly, RFR, PLSR, and HyFIS models in terms of robustness for biophysical and biochemical parameters retrieval using hyperspectral data.

Robustness of neural network calibration model for accurate spatial positioning.

The present study devotes to a systematical exploration for the robustness of neural network-based camera calibration method in the circumstance of three-dimensional (3D) spatial positioning via machine vision technique. By analyzing the error propagation route in the calibration-reconstruction process, a dimensionless error attenuation coefficient is proposed to measure the robustness of a calibration model with respect to input calibration error. Using this metric, the robustness of the neural network (NN) model under different optical configurations, i.e., input noise level, optical distortion and camera viewing angle, are analyzed in detail via synthetic simulation. Due to its generalized fitting capacity, the NN model is found to be superior to conventional pinhole model and polynomial model in terms of model robustness. To take full advantage of this feature, the NN model is further deployed to the scenarios of asymmetric camera layout and multiple camera joint calibration. Both synthetic simulation and experiment test demonstrate that the NN model can significantly improve the robustness and the accuracy of 3D spatial positioning in these non-normal scenarios.

The superiority of the normalized difference phenology index (NDPI) for estimating grassland aboveground fresh biomass

Accurate monitoring of grassland aboveground fresh biomass (called AGB in the study) and its spatial-temporal dynamics is indispensable for sustainable grassland management. The most common method used in estimating AGB with remotely sensed data is based on the relationship between field AGB measurements and vegetation indices (VIs); however, the existing VIs do not deliver adequate results due to the soil background and spatial, temporal and sampling size variability. In this study, the AGB estimation model with the normalized difference phenology index (NDPI) was evaluated in terms of model robustness and spatial and temporal scalability based on comparisons with the widely used ratio vegetation index (RVI), difference vegetation index (DVI), normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), soil-adjusted vegetation index (SAVI), modified soil-adjusted vegetation index (MSAVI), and optimized soil-adjusted vegetation index (OSAVI). The field measurements of AGB of the natural grassland in Inner Mongolia, China, collected in 2013, 2016, and 2017 and Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance products were used for analysis. The results based on training and independent validation data showed the following: (1) The R2 value between AGB and the NDPI was the highest (0.73) among all VIs, followed by soil-line-adjusted VIs, while the R2 values of the RVI and DVI were the lowest; (2) The NDPI-based model had the best robustness for different sampling sizes; (3) The NDPI-based model also had superior spatial and temporal scalability. The results from simulation experiments using the PROSAIL model also support the superiority of the NDPI in estimating AGB. The simulation analysis further reveals that the overall superiority of the NDPI originates from the fact that the NDPI overcomes the adverse impacts of the heterogeneity of the soil background and accounts for changes in the leaf water content that contribute substantially to AGB in grassland. These findings suggest that the NDPI-based AGB estimation model is advantageous for monitoring AGB in large grasslands with significant spatial-temporal heterogeneity.

Efficient Algorithms for E-Healthcare to Solve Multiobject Fuse Detection Problem

Object detection plays a vital role in the fields of computer vision, machine learning, and artificial intelligence applications (such as FUSE-AI (E-healthcare MRI scan), face detection, people counting, and vehicle detection) to identify good and defective food products. In the field of artificial intelligence, target detection has been at its peak, but when it comes to detecting multiple targets in a single image or video file, there are indeed challenges. This article focuses on the improved K-nearest neighbor (MK-NN) algorithm for electronic medical care to realize intelligent medical services and applications. We introduced modifications to improve the efficiency of MK-NN, and a comparative analysis was performed to determine the best fuse target detection algorithm based on robustness, accuracy, and computational time. The comparative analysis is performed using four algorithms, namely, MK-NN, traditional K-NN, convolutional neural network, and backpropagation. Experimental results show that the improved K-NN algorithm is the best model in terms of robustness, accuracy, and computational time.

The multivariate tail-inflated normal distribution and its application in finance

The research objective of this paper is to handle situations where the empirical distribution of multivariate real-valued data is elliptical and with heavy tails. Many statistical models already exist that accommodate these peculiarities. This paper enriches this branch of literature by introducing the multivariate tail-inflated normal (MTIN) distribution, an elliptical heavy tails generalization of the multivariate normal (MN). The MTIN belongs to the family of MN scale mixtures by choosing a convenient continuous uniform as mixing distribution. Moreover, it has a closed-form for the probability density function characterized by only one additional ‘inflation’ parameter, with respect to the nested MN, governing the tail-weight. The moment generating function, and the first four moments, are also derived; interestingly, the latter always exist and the excess kurtosis can assume any positive value. The method of moments and maximum likelihood (ML) are considered for estimation. As concerns the latter, a direct approach, as well as a variant of the EM algorithm (namely, the ECME algorithm), are illustrated. Furthermore, a way to approximate covariance matrix of the ML estimator is suggested and the existence of the ML estimates is evaluated. Since the inflation parameter is estimated from the data, robust estimates of the mean vector of the nested MN distribution are automatically obtained by down-weighting. Simulations are performed to compare the estimation methods/algorithms, to investigate the ability of AIC and BIC to select among a set of candidate elliptical models, and to evaluate the robustness of these candidate methods when data are skewed. The findings are the following: ML is better than MM, direct ML is suggested for low dimensions, while the ECME algorithm is to be preferred when the number of variables is higher, AIC and BIC work comparably in selecting the true underlying model, and the MTIN outperforms the competing models in terms of robustness toward skew data. For illustrative purposes, the MTIN distribution is finally fitted to multivariate financial data and compared with other well-established multivariate elliptical distributions. The analysis shows how the proposed model represents a valid alternative to the considered competitors in terms of AIC and BIC, but also in reproducing the higher empirical kurtosis which is common in the financial context.

A new binary level set model using L0 regularizer for image segmentation

In this paper, a new model of image segmentation is proposed by considering L0 regularizer term. A new energy function is formulated by utilizing Local Gaussian Distribution model based on binary level set function followed by introducing a L0 gradient regularizer as regularizing term. Instead of zero level set, the binary level set function is applied to differentiate between foreground and background regions. The regularization function is used to calculate the interfaces between foreground sub-regions and regularization term L0 which helps to evolve the curve. The proposed energy function is solved using minimization algorithm to achieve promising results in terms of segmentation accuracy. The different sets of experiments are performed on real and medical images. The proposed model provides higher segmentation accuracy results in less computational time compared to the other state-of-the-art models. Further, the results are found to be superior as compared to other existing models in terms of robustness to noise and intensity inhomogeneity.

Multi-criteria comprehensive study on predictive algorithm of heating energy consumption of district heating station based on timeseries processing

Refined control of district heating station relies on reasonable and accurate prediction of heating energy consumption. Due to the influence of building thermal inertia and time-delay of the district heating system, certain research work is necessary to thoroughly illustrate and analyze the impact of data timeseries processing on various prediction models. Four kinds of prediction algorithms were investigated and compared in this paper. Results showed that all of three timeseries processing methods, namely time feature construction, sliding window and building thermal inertia coefficient (CBTI), can improve the prediction accuracy of all four models and CBTI had the greatest impact on model accuracy improvement. Moreover, timeseries processing method has no limitation on the types of prediction model and it is a general method to improve the model accuracy. Time feature construction and sliding window had greater influence on the non-neural network models while CBTI was the opposite. In terms of model robustness, the robustness had been significantly improved after introducing timeseries processing except random forest, and the comprehensive robustness coefficient for the other three models had been reduced by about 95%. Recurrent neural network had extremely excellent robustness under different temporal granularity.

A Dynamic Robust Restoration Framework for Unbalanced Power Distribution Networks

The increasing penetration of photovoltaic (PV) generators has led to a reduction in the effectiveness of existing strategies for restoring the power distribution network. This article proposes a dynamic robust restoration (DRR) framework for the recovery of outage power considering uncertain PV outputs and demands. This framework is presented in two subsequent steps. In the first step, optimal decisions regarding the network configurations are generated. The second step then computes the modified dynamic Distflow equations and constraints under consideration of the worst operating conditions over the associated uncertainty sets with the aim of maximizing the recovery of outage power. The DRR model is formulated as a bilevel mixed-integer linear programming problem. A decomposition algorithm in a master–sub structure is used to solve the resulting system. The results of case studies show that the proposed DRR model yields obvious advantages over the existing deterministic dynamic restoration model in terms of robustness against system uncertainties.

Radar Signal Modulation Recognition Based on Deep Joint Learning

The development of integrated avionics systems and electromagnetic spectrum technology has attracted widespread attention. It has further increased the performance requirements for modulation signal recognition technology in complex electromagnetic environments. Therefore, this paper proposes a deep joint learning technique, including deep representation and low-dimensionality discrimination, to enhance feature stability and environmental adaptability. Specifically, we design a feature learning network based on AlexNet to extract in-depth features and optimize it through parameter-based transfer learning techniques, promote multi-level representation capabilities of features and reduce the sample size requirements. Moreover, we propose a classification algorithm based on kernel collaborative representation and discriminative projection to enhance the ability of low-dimensionality representation and between-class discrimination, which optimized using the mini-batch random gradient descent method. As shown in the simulation, the overall average recognition success rate of this method aiming at twelve radar signal modulation types reaches 97.58% at SNR of −6dB. The results of simulation and analysis demonstrate the superiority of the proposed model in terms of robustness, timeliness, and adaptability to small samples.

How meaningful are similarities in deep trajectory representations?

Finding similar trajectories is an important task in moving object databases. However, classical similarity models face several limitations, including scalability and robustness. Recently, an approach named t2vec proposed transforming trajectories into points in a high dimensional vector space, and this transformation approximately keeps distances between trajectories. t2vec overcomes that scalability limitation: Now it is possible to cluster millions of trajectories. However, the semantics of the learned similarity values – and whether they are meaningful – is an open issue. One can ask: How does the configuration of t2vec affect the similarity values of trajectories? Is the notion of similarity in t2vec similar, different, or even superior to existing models? As for any neural-network-based approach, inspecting the network does not help to answer these questions. So the problem we address in this paper is how to assess the meaningfulness of similarity in deep trajectory representations. Our solution is a methodology based on a set of well-defined, systematic experiments. We compare t2vec to classical models in terms of robustness and their semantics of similarity, using two real-world datasets. We give recommendations which model to use in possible application scenarios and use cases. We conclude that using t2vec in combination with classical models may be the best way to identify similar trajectories. Finally, to foster scientific advancement, we give the public access to all trained t2vec models and experiment scripts. To our knowledge, this is the biggest collection of its kind.

Robust Routing Optimization for Smart Grids Considering Cyber-Physical Interdependence

A smart grid is a typical cyber-physical system (CPS). Cyber networks and physical networks of smart grids have similar topologies and interdependent characteristics, which may induce cyber-physical coupling failures. Remedial control for physical outages may fail due to a simultaneous cyber-side failure, thereby increasing the risk to smart grids. To improve the robustness of smart grids in the face of possible cyber-physical coupling failures, the critical information flow should be allocated to reliable paths to ensure accessibility. The existing routing for power communication networks follows the information flow fairness principle. However, the information flow has different levels of importance in different power system states. In this paper, an importance evaluation approach for information flow based on the cyber-physical sensitivity is introduced. Then, a CPS robust routing model (CPS-RRM) with a priority mechanism that considers cyber-physical disturbances is proposed based on robust optimization. The formulated CPS-RRM is converted to a linear problem using the Big-M method and is solved using a modified C-CG algorithm. The superiority of our model is validated by comparing it to conventional routing based on the shortest-path model in terms of robustness and performance in power flow corrective control.