Unknown Shift Research Articles

A new omnibus SPRT chart for monitoring process mean and variability based on the average number of observations to signal

The recent development of the omnibus sequential probability ratio test (OSPRT) chart marks a significant contribution to the advancement of joint monitoring schemes. As the OSPRT chart is a variable-sample-size control chart, practitioners often wish to understand its inspection efficiency, i.e. the number of observations it samples before producing a signal. In this article, we propose two enhanced optimization designs for the OSPRT chart based on the average number of observations of signal (ANOS) and expected value of the ANOS (EANOS) metrics under deterministic and unknown shift sizes, respectively. The ANOS metric is central to our design as it perfectly combines both the average run length (ARL) and the average sample number. A comparative analysis reveals that the OSPRT chart outperforms four benchmarking control charts in terms of the ANOS and EANOS metrics. Finally, an implementation of the OSPRT chart is presented with a ball shear test dataset.

A distribution-free adaptive CUSUM-sign chart for monitoring shifts in the location of unknown industrial process

In most of the manufacturing industry, the distributional assumption of the online working processes is customarily unknown or hard to meet. In such situations, parametric charting mechanisms generally designed under normality assumptions of the model yield more false alarms and invalid out-of-control (OOC) comparisons. The nonparametric (distribution-free) charts are a better choice for practitioners in such cases as their in-control (IC) run length (RL) profiles remain the same. The study intends to develop a new distribution-free adaptive CUSUM sign (NPACUSUM-SN; named hereafter) chart for monitoring the process location. The proposed NPACUSUM-SN chart estimates the unknown process mean shift using an unbiased function and updates adaptively the reference parameter of the CUSUM statistic. The IC robustness of the conventional adaptive CUSUM and the proposed NPACUSUM-SN charts under the symmetric, skewed, and contaminated normal (CN) distributions have been computed using Monte Carlo simulations. The OOC RL profiles of the proposed study have been assessed for the initial state and the shift delay (steady state) in the processes. The proposed NPACUSUM-SN chart shows more resistance against non-normality and effective behaviour as compared to its conventional competitor. The proposed NPACUSUM-SN chart provides uniformly efficient RL characteristics as compared to its counterparts. Implementation of the proposal is made by using a manufacturing industry dataset along with an artificial dataset.

Dynamic Quantification With Constrained Error Under Unknown General Dataset Shift

Dynamic Quantification With Constrained Error Under Unknown General Dataset Shift

A variable sampling interval EWMA control chart for urinary tract infections data

Control chart is an important method in Statistical Process Control, which is used to study how a process changes over time. The exponential distribution is one of the commonly used models for fitting data, especially in areas such as component life, aviation accidents, or health-care processes. In order to further improve the sensitivity of the control chart, we propose an exponentially weighted moving average control chart with a variable sampling interval scheme, denoted as VSIEWMA-T control chart, for monitoring exponentially distributed quality characteristics (e.g. Urinary Tract Infections data). In this paper, the average time to signal property of the proposed control chart is investigated using the Markov chain method. In addition, parameter optimization algorithms for known or unknown shift levels are provided. Subsequently, numerical analysis shows that the proposed control chart outperforms other competitive control charts. Finally, an example of urinary tract infections data is provided for illustration.

Chemical fault diagnosis network based on single domain generalization

Recent developments in fault diagnosis have leveraged domain generalization to address the issue of domain shift. Most existing methods focus on learning domain-invariant representations from multiple source domains. However, collecting valuable fault samples from varying operational conditions is challenging, and it is common for available data to originate from a single operational condition. Thus, this paper introduces a Multi-scale generative and adversarial Metric networks (MGAMN) for Chemical Process Fault Diagnosis. To enhance model generalization, a domain generation module was developed to create augmented domains with significant distributional differences from the source domain. The diagnostic task module then extracts domain-invariant features from both the source and augmented domains. A multi-scale generation strategy is established, utilizing multi-scale deep separable convolutions (Dsc) to ensure that the generated samples contain rich state information. Additionally, an adversarial training and metric learning strategy is designed to learn generalized features capable of resisting unknown domain shifts. Extensive diagnostic experiments on the non-isothermal continuous stirred tank reactor (CSTR) and the Tennessee Eastman Process (TEP) chemical datasets validate the effectiveness of the proposed method. Moreover, ablation studies confirm the effectiveness of the proposed modular strategy, demonstrating significant potential for practical applications.

AdvST: Revisiting Data Augmentations for Single Domain Generalization

Single domain generalization (SDG) aims to train a robust model against unknown target domain shifts using data from a single source domain. Data augmentation has been proven an effective approach to SDG. However, the utility of standard augmentations, such as translate, or invert, has not been fully exploited in SDG; practically, these augmentations are used as a part of a data preprocessing procedure. Although it is intuitive to use many such augmentations to boost the robustness of a model to out-of-distribution domain shifts, we lack a principled approach to harvest the benefit brought from multiple these augmentations. Here, we conceptualize standard data augmentations with learnable parameters as semantics transformations that can manipulate certain semantics of a sample, such as the geometry or color of an image. Then, we propose Adversarial learning with Semantics Transformations (AdvST) that augments the source domain data with semantics transformations and learns a robust model with the augmented data. We theoretically show that AdvST essentially optimizes a distributionally robust optimization objective defined on a set of semantics distributions induced by the parameters of semantics transformations. We demonstrate that AdvST can produce samples that expand the coverage on target domain data. Compared with the state-of-the-art methods, AdvST, despite being a simple method, is surprisingly competitive and achieves the best average SDG performance on the Digits, PACS, and DomainNet datasets. Our code is available at https://github.com/gtzheng/AdvST.

"Please may I have some more?" – Assessing Student and Instructor Intersubjective Agreement on Part Marks in Engineering Mathematics

Driven by student requests for more marks (some prefer ‘partial credit’), this paper asks if there is evidence that students adapt to the marking rubrics encountered in universities. Through a weekly handout in a first-year engineering mathematics class, we explore the progress students make in their understanding of how instructors grade. How does student expectation of an unknown grading rubric shift with the accumulation of experience in a course? Through our exploration, we reveal unfortunate findings about how well students adapt, a lot of unanswered questions, a few sources of good news, and promising directions for the research to pursue.

Contrast-Assisted Domain-Specificity-Removal Network for Semi-Supervised Generalization Fault Diagnosis.

Unknown domain shift caused by the unavailability of target domain during training phase degrades the performance of intelligent fault diagnosis models in practical applications. Domain generalization (DG)-based methods have recently emerged to alleviate the influence of domain shift and improve the generalization ability of models toward invisible working conditions. However, most existing studies are conducted on multiple fully labeled source domains. Meanwhile, domain-specific information related to the variations of working conditions is often neglected during model training. Therefore, in order to realize reliable generalization fault diagnosis based on partially labeled source domains, this article proposes a contrast-assisted domain-specificity-removal network (CDSRN) to extract transferable features from domain-specificity-removal perspective. Concretely, a domain-specific feature removal branch is designed to disentangle domain-invariant features and domain-specific features, thus excavating generalized information only in domain-invariance dimension. Simultaneously, proxy-contrastive representation enhancement module is embedded to facilitate the fault class-discriminative and domain-discriminative feature learning, thereby assisting the model in further improvement of generalization capability. Experimental studies confirm the effectiveness and competitiveness of the proposed CDSRN in semi-supervised generalization fault diagnosis.

Variable Sample Size Control Charts for Monitoring the Multivariate Coefficient of Variation Based on Median Run Length and Expected Median Run Length

The monitoring of a well-functioning process system has always held significant importance. In recent times, there has been notable attention towards employing control charts to oversee both univariate and multivariate coefficients of variation (MCV). This shift is in response to the concern of erroneous outcomes that can arise when traditional control charts are applied under the condition of dependent mean and standard deviation, as highlighted by prior research. To address this, the remedy lies in adopting the coefficient of variation. Furthermore, this study underscores the application of MCV in scenarios where multiple quality attributes are simultaneously under surveillance within an industrial process. This aspect has demonstrated considerable enhancement in chart performance, especially when incorporating the variable sample size (VSS) feature into the MCV chart. Adaptive VSS, evaluated through metrics like median run length (MRL) and expected median run length (EMRL), is also integrated for MCV monitoring. In contrast to earlier studies that predominantly focused on average run length (ARL), this research acknowledges the potential inaccuracies in ARL measurement. In this study, two optimal designs for VSS MCV charts are formulated by minimizing two criteria: firstly, MRL; and secondly, EMRL, both accounting for deterministic and unknown shift sizes. Additionally, to assess the distribution's variability in run lengths, the study provides the 5th and 95th percentiles. The research delves into two VSS schemes: one with a defined small sample size (nS), and another with a predetermined large sample size (nL) for the initial subgroup (n(1)). The approach taken involves the development of a Markov chain method for designing and deriving performance measures of the proposed chart. These measures include MRL and EMRL. Moreover, a comparative analysis between the proposed chart's performance and the standard MCV chart (STD) is presented in terms of MRL and EMRL criteria. The outcomes illustrate the superiority of the proposed chart over the STD MCV chart for all shift sizes, whether they are upward or downward, and when n(1) equals nS or nL.

Distributionally Robust Learning With Stable Adversarial Training

Machine learning algorithms with empirical risk minimization are vulnerable under distributional shifts due to the greedy adoption of all the correlations found in training data. There is an emerging literature on tackling this problem by minimizing the worst-case risk over an uncertainty set. However, existing methods mostly construct ambiguity sets by treating all variables equally regardless of the stability of their correlations with the target, resulting in the overwhelmingly-large uncertainty set and low confidence of the learner. In this paper, we propose a novel Stable Adversarial Learning (SAL) algorithm that leverages heterogeneous data sources to construct a more practical uncertainty set and conduct differentiated robustness optimization, where covariates are differentiated according to the stability of their correlations with the target. We theoretically show that our method is tractable for stochastic gradient-based optimization and provide the performance guarantees for our method. Empirical studies on both simulation and real datasets validate the effectiveness of our method in terms of uniformly good performance across unknown distributional shifts.

An Expected Average Run Length (EARL) Performance Comparison of the SSGR and EWMA Control Charts

The acceleration use of control charts in industrial processes has led to the effectiveness in their evaluation by quality practitioners. This is crucial, as it influences their decisions on the choice of which control charts to employ. This study aims to explore and compare the performance of the side sensitive group runs (SSGR) and exponentially weighted moving average (EWMA) control charts. In general, the average run length (ARL) characteristics were used to evaluate the performance of these control charts. The ARL, which considers the exact shift size in the process, is restricted in the case when the practitioner cannot identify the process shift size (unknown shift size). In this situation, the expected average run length (EARL) is an alternative performance criterion. Upon comparison of the findings obtained, the EWMA chart has superior performance when (δmin, δmax) = (0.1, 0.4). In contrast, the SSGR chart overtakes the EWMA chart when (δmin, δmax) = (0.5, 0.8) and (δmin, δmax) = (0.9, 1.2), except when the sample size n = 3 for (δmin, δmax) = (0.5, 0.8). For this particular combination, the EWMA chart performs slightly better than the SSGR chart. The outcome of this study is expected to contribute to practitioners in identifying suitable control charts in process monitoring and implementation.

Application of Wavelet Transform to Urysohn-Type Equations

This paper deals with convolution-type Urysohn equations of the first kind. Finding a solution for such equations is an ill-posed problem. For it to be solved, regularization algorithms and the continuous wavelet transform are used. Similar to the Fourier transform, the continuous wavelet transform is applied to convolution-type equations (based on the Fourier and wavelet transforms) and to Urysohn equations with unknown shift. The wavelet transform is preferable for the cases with approximated right-hand sides and for type 1 equations. We demonstrated that the application of the wavelet transform to Urysohn-type equations with unknown shift translates into a solution of a nonlinear equation with an oscillating kernel. Depending on the availability of a priori information, a combination of regularization and iterative algorithms with the use of close equations are effective for solving convolution-type equations based on the continuous wavelet transform and Urysohn equation.

Survey of Point Cloud Registration Methods and New Statistical Approach

The use of a 3D range scanning device for autonomous object description or unknown environment mapping leads to the necessity of improving computer methods based on identical point pairs from different point clouds (so-called registration problem). The registration problem and three-dimensional transformation of coordinates still require further research. The paper attempts to guide the reader through the vast field of existing registration methods so that he can choose the appropriate approach for his particular problem. Furthermore, the article contains a regression method that enables the estimation of the covariance matrix of the transformation parameters and the calculation of the uncertainty of the estimated points. This makes it possible to extend existing registration methods with uncertainty estimates and to improve knowledge about the performed registration. The paper’s primary purpose is to present a survey of known methods and basic estimation theory concepts for the point cloud registration problem. The focus will be on the guiding principles of the estimation theory: ICP algorithm; Normal Distribution Transform; Feature-based registration; Iterative dual correspondences; Probabilistic iterative correspondence method; Point-based registration; Quadratic patches; Likelihood-field matching; Conditional random fields; Branch-and-bound registration; PointReg. The secondary purpose of this article is to show an innovative statistical model for this transformation problem. The new theory needs known covariance matrices of identical point coordinates. An unknown rotation matrix and shift vector have been estimated using a nonlinear regression model with nonlinear constraints. The paper ends with a relevant numerical example.

A variable sampling interval one-sided CUSUM control chart for monitoring the multivariate coefficient of variation

Monitoring techniques based on the multivariate coefficients of variation (MCV) have received a great deal of attention in quality control. Numerous studies have shown that adaptively changing the charting parameters based on the past sample information can improve the performance of a chart in detecting process changes. In view of the performance benefits of adaptive strategies, two one-sided cumulative sum (CUSUM) charts are proposed to monitor the MCV and a variable sampling interval (VSI) strategy is incorporated into these charts. By using the Markov chain method, the formulas for calculating the average and standard deviation of time to signal measures of the VSI CUSUM MCV control charts are derived. Then for the known and unknown shift sizes, the optimization algorithms for obtaining the charting parameters are presented. The superiority of the proposed control charts is confirmed by comparing their performance with two existing MCV charts. Finally, an example using real investment data demonstrates the practical application of the proposed charts.

Prediction sets adaptive to unknown covariate shift.

Predicting sets of outcomes-instead of unique outcomes-is a promising solution to uncertainty quantification in statistical learning. Despite a rich literature on constructing prediction sets with statistical guarantees, adapting to unknown covariate shift-a prevalent issue in practice-poses a serious unsolved challenge. In this article, we show that prediction sets with finite-sample coverage guarantee are uninformative and propose a novel flexible distribution-free method, PredSet-1Step, to efficiently construct prediction sets with an asymptotic coverage guarantee under unknown covariate shift. We formally show that our method is asymptotically probably approximately correct, having well-calibrated coverage error with high confidence for large samples. We illustrate that it achieves nominal coverage in a number of experiments and a data set concerning HIV risk prediction in a South African cohort study. Our theory hinges on a new bound for the convergence rate of the coverage of Wald confidence intervals based on general asymptotically linear estimators.

Adaptive attitude angle constrained fault-tolerant control of hypersonic vehicle with unknown centroid shift

This work takes a further step into the adaptive fault-tolerant control (FTC) scheme to cope with the challenges deriving from the hypersonic vehicle (HSV) in the presence of unknown centroid shift, input saturation, and actuator fault. The influence of unknown centroid shift is mainly manifested in the following aspects: 1) uncertainties of the system input matrix, 2) mixed system uncertainties, and 3)eccentric moments. A novel attitude state constraint control strategy is technically proposed by incorporating the extended barrier Lyapunov function (Ex-BLF) and the Nussbaum-type function into the backstepping design to keep the safe flight of HSV. A unified controller is designed to handle the cases with/without state constraints based on the Ex-BLF with its auxiliary system. For unknown mixed system input matrix due to the coupling of uncertainties of the inertial matrix, actuator fault, and system input saturation, a specific Nussbaum-type function assisted by a state-depend auxiliary system is designed to compensate for the influence of those time-varying nonlinear terms. As a consequence, combined with the Lyapunov stability theory, it is confirmed that the proposed FTC scheme ensures that all the closed-loop signals are bounded. Simulation results are carried out to illustrate the effectiveness and advantage of the proposed control scheme.

АЛГОРИТМ ОБНАРУЖЕНИЯ ФМ-СИГНАЛОВ НА ОСНОВЕ ВИНЕРОВСКОЙ ФИЛЬТРАЦИИ С ПРЕДВАРИТЕЛЬНОЙ НЕЛИНЕЙНОЙ ОБРАБОТКОЙ

In this paper, an algorithm for preprocessing of PSK - signals based on wiener filtration is proposed. The aim of preprocessing is to extract the phase discontinuities when filtering harmonic filling of a signal. This method allows us to avoid compensation of unknown frequency shift in the problems of determining the time delay of signals during multichannel propagation. The study checks the conditions of algorithm applicability and its stability to a noise level in the observed channels.

Estimating phase parameters of a three-level system interacting with two classical monochromatic fields in simultaneous and individual metrological strategies

Recently, the Hilbert–Schmidt speed, as a special class of quantum statistical speed, has been reported to improve interferometric phase in a single-parameter quantum estimation. Here, we test this concept in the multiparameter scenario where two laser phases are estimated in a theoretical model consisting of a three-level atom interacting with two classical monochromatic fields. When the atom is initially prepared in the lower bare state, taking into account the detuning parameters, we extract an exact analytical solution of the atomic density matrix in the case of a two-photon resonant transition. Further, we compare the performance of laser phase parameter estimation in individual and simultaneous metrological strategies and explore the role of quantum coherence in improving the efficiency of unknown multiphase shift estimation protocols. The obtained results show that the Hilbert–Schmidt speed detects the lower bound on the statistical estimation error as well as the optimal estimation regions, where its maximal corresponds to the maximal quantum Fisher information; further, the performance of simultaneous multiparameter estimation with individual estimation inevitably depends on the detuning parameters of the three-level atom. Aside from the quantum entanglement, the quantum coherence is also a crucial resource to improve the accuracy of a metrological protocol.

An optimally designed distribution‐free CUSUM procedure for tri‐aspect surveillance of continuous processes

AbstractIn the past few decades, research on nonparametric process monitoring schemes mainly dealt with the uni‐aspect or bi‐aspect schemes, focusing on monitoring process location or scale separately or jointly. Another critical process characteristic, namely, the process shape, is not explicitly dealt with in‐depth in most existing charting schemes. In classical hypothesis testing, some recent literature clearly showed that the multi‐aspect test statistics, although designed for very restrictive alternatives, often perform as well or better than many statistics for arbitrary distributional shifts. They are often better than Kolmogorov‐Smirnov or Cramér‐von Mises statistics and some empirical likelihood‐based test statistics. The current paper aims to use a tri‐aspect statistic to design a distribution‐free Phase‐II Cumulative Sum (CUSUM) charting scheme for monitoring any arbitrary process changes in the process. The proposal is nonparametric and is equivalent to an unknown standard case. It reflects, in addition, which parameters, among location, scale, or shape, are more responsible for a signal. The construction of the CUSUM scheme from an existing tri‐aspect Shewhart‐type chart is simple, so significant attention is devoted to determining a near‐optimal reference parameter of the chart in keeping an unknown shift type in mind. Comparisons of the optimal performance of various competitors are considered in terms of the median run length (MRL) metric. The proposed charting scheme designed with the tri‐aspect statistic compares highly favorably with many existing SPM schemes. The same is evident from our findings based on the Monte‐Carlo simulation. Finally, the proposed schemes are illustrated with a flow‐width measurement monitoring example.

Towards Accurate Reconstruction of 3D Scene Shape From A Single Monocular Image.

Despite significant progress made in the past few years, challenges remain for depth estimation using a single monocular image. First, it is nontrivial to train a metric-depth prediction model that can generalize well to diverse scenes mainly due to limited training data. Thus, researchers have built large-scale relative depth datasets that are much easier to collect. However, existing relative depth estimation models often fail to recover accurate 3D scene shapes due to the unknown depth shift caused by training with the relative depth data. We tackle this problem here and attempt to estimate accurate scene shapes by training on large-scale relative depth data, and estimating the depth shift. To do so, we propose a two-stage framework that first predicts depth up to an unknown scale and shift from a single monocular image, and then exploits 3D point cloud data to predict the depth shift and the camera's focal length that allow us to recover 3D scene shapes. As the two modules are trained separately, we do not need strictly paired training data. In addition, we propose an image-level normalized regression loss and a normal-based geometry loss to improve training with relative depth annotation. We test our depth model on nine unseen datasets and achieve state-of-the-art performance on zero-shot evaluation. Code is available at: https://github.com/aim-uofa/depth/.