Search

Measurement error always exists in quality control applications and may considerably affect the ability of control charts to detect an out-of-control situation. In this paper, we study the performance of the EWMA median chart using a Markov Chain approach with a linear covariate error model and a corrected formula for the distribution of the sample median. The performance is evaluated through the Average Run Length. Several figures and tables are presented and enumerated to show the statistical performance of the EWMA median control chart for different sources of measurement errors. The positive effect of taking multiple measurements for each item in a subgroup on the performance of the EWMA median chart is also investigated. An example illustrates the use of this study is introduced. The numerical simulation shows that the measurement errors have a negative influence on the proposed chart.

International audience; In many industrial manufacturing processes, the ratio of the variance to the mean of a quantity of interest is an important characteristic to ensure the quality of the processes. This ratio is called the coefficient of variation (CV). A lot of control charts have been designed for monitoring the CV of univariate quantity in the literature. However, the CV control charts for multivariate quantity have not received much attention yet. In this paper, we investigate a variable sampling interval (VSI) Shewhart control chart for monitoring multivariate CV. The time between two consecutive samples is allowed to vary according to the previous value of the multivariate CV, which will help the chart to detect the process shifts faster. The comparison with the fixed sampling interval Shewhart chart is implemented to highlight the advantage of the VSI method. Finally, an illustrative example is demonstrated on real data.

Online monitoring of the multivariate coefficient of variation (MCV) can be of interest in many real situations in which the dispersion of a multivariate process is meant to remain constant with regard to its position. To this aim, several control charts have been recently proposed in the literature. In this paper, the new one-sided adaptive charts to monitor the MCV are proposed. The chart applies a variable sampling interval (VSI) strategy on an exponentially weighted moving average (EWMA) scheme, aiming at benefiting from the known advantages of both approaches. Formulas to optimally determine the parameters of the chart are derived and presented. The proposed chart is shown to outperform competing charts under most circumstances. The presence of the measurement errors is considered to understand their effects on the charts. An illustrative example is also included.

Part 1: Artificial Intelligence Based Optimization Techniques for Demand-Driven Manufacturing; International audience; With the development of e-commerce, payment by credit card has become an essential means for the purchases of goods and services online. Especially, the Manufacturing Sector faces a high risk of fraud online payment. Its high turnover is the reason making this sector is lucrative with fraud. This gave rise to fraudulent activity on the accounts of private users, banks, and other services. For this reason, in recent years, many studies have been carried out using machine learning techniques to detect and block fraudulent transactions. This article aims to present a new approach based on real-time data combining two methods for the detection of credit card fraud. We first use the variational autoencoder(VAE) to obtain representations of normal transactions, and then we train a support vector data description (SVDD) model with these representations. The advantage of the representation learned automatically by the variational autoencoder is that it makes the data smoother, which makes it possible to increase the detection performance of one-class classification methods. The performance evaluation of the proposed model is done on real data from European credit cardholders. Our experiments show that our approach has obtained good results with a very high fraud detection rate.

International audience; Among the anomaly detection methods, control charts have been considered important techniques. In practice, however, even under the normal behaviour of the data, the standard deviation of the sequence is not stable. In such cases, the coefficient of variation (CV) is a more appropriate measure for assessing system stability. In this paper, we consider the statistical design of Run Rules-based control charts for monitoring the CV of multivariate data. A Markov chain approach is used to evaluate the statistical performance of the proposed charts. The computational results show that the Run Rules-based charts outperform the standard Shewhart control chart significantly. Moreover, by choosing an appropriate scheme, the Run Rules-based charts perform better than the Run Sum control chart for monitoring the multivariate CV.

International audience; With a tremendous increase in mobile and wearable devices, the study of sensor-based activity recognition has drawn a lot of attention in the past years. In recent years, the applications of Human Activity Recognition are getting more and more attention, especially in eldercare and healthcare as an assistive technology when combined with the Internet of Things. In this paper, we propose three deep learning approaches to improve the accuracy of activity detection on the WISDM dataset. Particularly, we apply a convolutional neural network to extract the interesting features, then we use softmax function, support vector machine, and random forest for classification tasks. The results show that the hybrid algorithm, convolutional neural network combined with the support vector machine, outperforms all the previous methods in classifying every activity. In addition, not only the support vector machine but also the random forest shows better accuracy in classification tasks than the neural network classification and the former approaches do.

International audience; We study the effect of nutritional diet characteristics on the lactating Holstein-Friesian dairy cows in Brittany, France from 36 individuals. An analysis ofthe relations between fat/protein content and milk yield was implemented for our dataset. The fat and protein production increase at a slower rateas milk yield increases. The importance of chemical composition on milk production is studied using the linear model. The data analysis confirms theimportance of Starch, crude fiber, and protein which have a positive effect on milk production. This analysis also confirms the previous study on the effect of parity on the production. After that, the milk production forecasting is investigated using both linear models and machine learning approaches (support vector machine, random forest, neural network). We study the performance of multiple linear regression and machine learning-based models in both non-autoregressive and autoregressive cases at the individual level. The autoregressive models, which take into account the previously observed milk yield, have proven to significantly outperform the non-autoregressive approaches. Moreover, the computational cost of each approach is presented in the paper. While the random forest algorithm gives the best performance in both non-autoregressive and autoregressive approaches. The support vector machine algorithm gives a very close performance with a substantial less computing time. The support vector machine is shown to be the best compromise between accuracy and computational cost.

Identifying a forged printed document with scanned evidence can be a challenge. Microscopic printing is showing random shape which depends on the printing source as well as printing material. This paper presents a statistical analysis of the printing patterns under a microscopic scale, analyses the effect of printing direction, printing substrate (uncoated and coated paper), and printing technology (conventional offset, waterless offset, and electrophotography). The analysis shows a negligible effect of printing direction, yet, using the shape descriptor indexes, the printing materials and technologies are distinguishable under a microscopic scale. As a result, the algorithms based on Support Vectors Machine and Random Forest are developed, with shape descriptor indexes as features, for printing source identification. Both proposed algorithms, equally, achieve a high classification accuracy rate, over 92% accuracy with complex geometric-shape patterns. Thanks to the lightweight and efficiency of the Support Vectors Machine, the study shows promising applications for real-world and potential implementation in the Internet of Things devices.

International audience; In this study, we propose a new approach to determine intrusions of network in real-time based on statistical process control technique and kernel null space method. The training samples in a class are mapped to a single point using the Kernel Null Foley-Sammon Transform. The Novelty Score are computed from testing samples in order to determine the threshold for the real-time detection of anomaly. The efficiency of the proposed method is illustrated over the KDD99 data set. The experimental results show that our new method outperforms the OCSVM and the original Kernel Null Space method by 1.53% and 3.86% respectively in terms of accuracy.

International audience; Shewhart's type control charts for monitoring the Multivariate Coefficient of Variation (MCV) have recently been proposed in order to monitor the relative variability compared with the mean. These approaches are known to be rather slow in the detection of small or moderate process shifts. In this paper, in order to improve the detection efficiency, two one-sided Synthetic charts for the MCV are proposed. A Markov chain method is used to evaluate the statistical performance of the proposed charts. Furthermore, computational experiments reveal that the proposed control charts outperform the Shewhart MCV control chart in terms of the average run length to detect an out-of-control state. Finally, the implementation of the proposed chart is illustrated with an example using steel sleeves data.

International audience; In this paper, we propose a variable sampling interval Shewhart control chart to monitor the coefficient of variation (CV) squared, denoted by VSI SH-γ 2. The new model overcomes the ARL-biased (average run length) property of the control chart monitoring the CV in a previous study by designing two one-sided charts rather than one two-sided chart. Moreover, the effect of measurement error on the performance of the VSI SH-γ 2 control chart is investigated. The incorrect formula for the distribution of the CV in the presence of measurement error in a former study is fixed. Numerical simulations show that the precision errors and accuracy errors do have negative influences on the VSI SH-γ 2 chart. An appropriate strategy based on the obtained results is suggested to reduce these negative effects.

International audience; In this paper, we present a method to monitor the coefficient of variation (CV) squared using two one-sided synthetic control charts. The numerical results show that our design outperforms the two-sided synthetic control chart monitoring the CV. The steady-state, which is have practical meaning in many situations, is also considered. We use a Markov chain method to evaluate the statistical performance of the proposed charts. Furthermore, the effect of measurement errors on synthetic control charts monitoring the CV squared is firstly investigated.

International audience; One-class support vector machines (OCSVM) have been recently applied in intrusion detection. Typically, OCSVM is kernelized by radial basis functions (RBF, or Gaussian kernel) whereas selecting Gaussian kernel hyperparameter is based upon availability of attacks, which is rarely applicable in practice. This paper investigates the application of nested OCSVM to detect intruders in network systems with data-driven hyperparameter optimization. The nested OCSVM is able to improve the efficiency over the proposed OCSVM applied in intrusion detection. In addition , the information of the farthest and the nearest neighbors of each sample is used to construct the objective cost instead of labeling based metrics such as geometric mean accuracy. The efficiency of this method is illustrated over the KDD99 dataset whereas the resulting estimated boundary, as well as intrusion detection performance, are comparable with existing methods. The experimental results show that the nested OCSVM method performs better than OCSVM for intrusion detection. The nested OCSVM with 12 density levels achieves 98.28% in accuracy and higher true alarming rate (TP) comparing to OCSVM.

International audience; Microscopic analysis of paper printing shows regularly spaced dots whose random shape depends on the printing technology, the configuration of the printer as well as the paper properties. The modelling and identification of paper and ink interactions are required for qualifying the printing quality, for controlling the printing process and for application in authentication as well. This paper proposes an approach to identify the authentic printer source using micro-tags consisting of microscopic printed dots embedded in the documents. These random shape features are modelled and extracted as a signature for a particular printer. In the paper, we propose a probabilistic model consisting of vector parameters using a spatial interaction binary model with inhomogeneous Markov chain. These parameters determine the location and describe the diverse micro random structures of microscopic printed dots. A Markov chain Monte Carlo (MCMC) algorithm is thus developed to approximate the Minimum Mean Squared Error estimator. The performance is assessed through numerical simulations. The real printed dots from the common printing technologies (conventional offset, waterless offset, inkjet, laser) are used to assess the effectiveness of the model.

International audience; In this paper we are concerned by authentication of printer technologies from microscopic analysis of paper print. At this scale, a print is made of regularly spaced dots whose shape varies from a print to another and also inside the same document. Thus, dot at the microscopic scale can be considered as an intrinsic signature of printer technologies. Modeling and estimating such a signature for the authentication of printer technologies are really challenging. In this paper, we propose an original modeling of the micrometric scan of document printing. It consists in an extension of the binary response model which takes into account the dot shape. The digital image of a dot is therefore modeled as a set of random pixels distributed following to the so called inverse link function which depends on the center, tone of black, its spreading and its shape. A maximum likelihood estimation algorithm is provided in order to estimate the location, the darkness, the scale and shape parameter of the dot. From experimental results on three different printer technologies (inkjet, laser and offset), we show that the shape parameter is relevant for designing an identification scheme of printer technologies.