Modified Expectation Maximization Research Articles

Reliability modeling for perception systems in autonomous vehicles: A recursive event-triggering point process approach

Ensuring the reliability of sensor-fusion-based perception systems is crucial for the safe deployment of autonomous vehicles. Such systems function through a sequence of interconnected stages, where errors in upstream stages may propagate to downstream stages and trigger additional errors. The cross-stage error propagation conceptually exists and makes errors in different stages, not independent, posing model challenges, estimation challenges, and data challenges for reliability modeling. The existing methods cannot be applied to address all these challenges. Thus, this paper presents a recursive event-triggering point process to explicitly consider the error propagation based on the simulated data. The data are simulated from a proposed error injection framework, which can generate various errors from a sequence of interconnected stages in a perception system. The latent and probabilistic error propagation information is incorporated into a modified expectation–maximization (EM) algorithm for parameter estimation. The numerical and physics-based simulation case studies demonstrate the prediction accuracy and interpretability of the proposed modeling methodology.

Advanced OCTA imaging segmentation: Unsupervised, non-linear retinal vessel detection using modified self-organizing maps and joint MGRF modeling

Background and Objective:This paper proposes a fully automated and unsupervised stochastic segmentation approach using two-level joint Markov-Gibbs Random Field (MGRF) to detect the vascular system from retinal Optical Coherence Tomography Angiography (OCTA) images, which is a critical step in developing Computer-Aided Diagnosis (CAD) systems for detecting retinal diseases. Methods:Using a new probabilistic model based on a Linear Combination of Discrete Gaussian (LCDG), the first level models the appearance of OCTA images and their spatially smoothed images. The parameters of the LCDG model are estimated using a modified Expectation Maximization (EM) algorithm. The second level models the maps of OCTA images, including the vascular system and other retina tissues, using MGRF with analytically estimated parameters from the input images. The proposed segmentation approach employs modified self-organizing maps as a MAP-based optimizer maximizing the joint likelihood and handles the Joint MGRF model in a new, unsupervised way. This approach deviates from traditional stochastic optimization approaches and leverages non-linear optimization to achieve more accurate segmentation results. Results:The proposed segmentation framework is evaluated quantitatively on a dataset of 204 subjects. Achieving 0.92 ± 0.03 Dice similarity coefficient, 0.69 ± 0.25 95-percentile bidirectional Hausdorff distance, and 0.93 ± 0.03 accuracy, confirms the superior performance of the proposed approach. Conclusions:The conclusions drawn from the study highlight the superior performance of the proposed unsupervised and fully automated segmentation approach in detecting the vascular system from OCTA images. This approach not only deviates from traditional methods but also achieves more accurate segmentation results, demonstrating its potential in aiding the development of CAD systems for detecting retinal diseases.

Stable training of probabilistic models using the leave-one-out maximum log-likelihood objective

Probabilistic modelling of power systems operation and planning processes depends on data-driven methods, which require sufficiently large datasets. When historical data lacks this, it is desired to model the underlying data generation mechanism as a probability distribution to assess the data quality and generate more data, if needed. Kernel density estimation (KDE) based models are popular choices for this task, but they fail to adapt to data regions with varying densities. In this paper, an adaptive KDE model is employed to circumvent this, where each kernel in the model has an individual bandwidth. The leave-one-out maximum log-likelihood (LOO-MLL) criterion is proposed to prevent the singular solutions that the regular MLL criterion gives rise to, and it is proven that LOO-MLL prevents these. Relying on this guaranteed robustness, the model is extended by adjustable weights for the kernels. In addition, a modified expectation–maximization algorithm is employed to accelerate the optimization speed reliably. The performance of the proposed method and models are exhibited on two power systems datasets using different statistical tests and by comparison with Gaussian mixture models. Results show that the proposed models have promising performance, in addition to their singularity prevention guarantees.

DINA Model with Entropy Penalization

The cognitive diagnosis model (CDM) is an effective statistical tool for extracting the discrete attributes of individuals based on their responses to diagnostic tests. When dealing with cases that involve small sample sizes or highly correlated attributes, not all attribute profiles may be present. The standard method, which accounts for all attribute profiles, not only increases the complexity of the model but also complicates the calculation. Thus, it is important to identify the empty attribute profiles. This paper proposes an entropy-penalized likelihood method to eliminate the empty attribute profiles. In addition, the relation between attribute profiles and the parameter space of item parameters is discussed, and two modified expectation–maximization (EM) algorithms are designed to estimate the model parameters. Simulations are conducted to demonstrate the performance of the proposed method, and a real data application based on the fraction–subtraction data is presented to showcase the practical implications of the proposed method.

Improved graph neural network-based green anaconda optimization for segmenting and classifying the lung cancer.

Normal lung cells incur genetic damage over time, which causes unchecked cell growth and ultimately leads to lung cancer. Nearly 85% of lung cancer cases are caused by smoking, but there exists factual evidence that beta-carotene supplements and arsenic in water may raise the risk of developing the illness. Asbestos, polycyclic aromatic hydrocarbons, arsenic, radon gas, nickel, chromium and hereditary factors represent various lung cancer-causing agents. Therefore, deep learning approaches are employed to quicken the crucial procedure of diagnosing lung cancer. The effectiveness of these methods has increased when used to examine cancer histopathology slides. Initially, the data is gathered from the standard benchmark dataset. Further, the pre-processing of the collected images is accomplished using the Gabor filter method. The segmentation of these pre-processed images is done through the modified expectation maximization (MEM) algorithm method. Next, using the histogram of oriented gradient (HOG) scheme, the features are extracted from these segmented images. Finally, the classification of lung cancer is performed by the improved graph neural network (IGNN), where the parameter optimization of graph neural network (GNN) is done by the green anaconda optimization (GAO) algorithm in order to derive the accuracy maximization as the major objective function. This IGNN classifies lung cancer into normal, adeno carcinoma and squamous cell carcinoma as the final output. On comparison with existing methods with respect to distinct performance measures, the simulation findings reveal the betterment of the introduced method.

CRF-MEM: Conditional Random Field Model Based Modified Expectation Maximization Algorithm for Sarcasm Detection in Social Media

<p>Text processing is an important task in various machine learning applications. One among the applications is Sentiment analysis. However, the presence of sarcasm makes it difficult for analyzing the sentiment of the statement. In the current scenario, the amount of sarcastic statements in any social media platform is high taking the forms of memes, comments, trolls etc. Hence it is important to identify sarcasm to preserve the polarity of any given statement. Sarcasm usually means the opposite of what the sentence seems to convey. While the existing works in literature have focused on detecting sarcasm, the proposed model, in addition to that, determines the levels of sarcasm present in the text, which will aid in finding the level of harshness present in the statement. In this work, an unsupervised learning model, Conditional Random Field model based Modified Expectation Maximization (CRF-MEM) algorithm has been proposed for detecting sarcasm in tweets. The proposed model aims to overcome the limitation present in the traditional EM algorithm, the random assignment factor, with the proposed aspect relationship value. Experimental results showed that the proposed CRF-MEM achieved an accuracy of 91.89% whereas the traditional EM displayed an accuracy of 80% in detecting sarcasm from text.</p> <p> </p>

A latent class Cox model for heterogeneous time-to-event data

Credit risk plays a vital role in the era of digital finance and it is one of primary interests to identify customers with similar types of risk categories so that personalized financial services can be offered accordingly. Motivated by the bourgeoning need for default risk modeling in finance, we propose herein a latent class Cox model for heterogeneous time-to-event data. The proposed model naturally extends the Cox proportional hazards model to flexibly take into account the heterogeneity of covariate effects as often manifested in real data. Without a priori specification of the number of latent classes, it simultaneously incorporates the commonalities and disparities of individual customers’ risk behaviors and provides a more refined modeling technique than existing approaches. We further propose a penalized maximum likelihood approach to identify the number of latent classes and estimate the model parameters. A modified expectation–maximization algorithm is then developed for its numerical implementation. Simulation studies are conducted to assess the finite-sample performance of the proposed approach. Its illustration with a real credit card data set is also provided.

Statistical inference on group Rasch mixture network models

In a two‐mode network, the nodes are divided into two types (primary nodes and secondary nodes), and connections exist only between nodes of different types. In reality, in such a two‐mode network, one‐mode network connections may also exist among primary nodes, and these two kinds of networks are usually not independent and coexistent. In this paper, we first propose a group Rasch mixture network model that focuses on the connections between primary nodes and secondary nodes, while incorporating the group structure and linkage information of primary nodes. We then develop a modified expectation–maximization algorithm to estimate the proposed model with a λ‐BIC method for selecting the tuning parameter. Additionally, we provide a likelihood‐ratio test statistic to examine whether the two kinds of networks are independent and implement the leave‐one‐out method to construct a network prediction rule. Finally, we establish asymptotic results and demonstrate the numerical performance of the proposed methods using both simulations and the Last.fm dataset.

Robust parameter estimation with outlier-contaminated correlated measurements and applications to aerodynamic coefficient identification

We consider the robust parameter estimation problem for state-space models with correlated measurements in the presence of outliers. Both the scenarios that the statistical information of the nominal noises in state-space models is known or not are explored. A novel Kalman robust smoother is proposed via introducing a specific reweighting approach to estimate the system parameters as well as the states when the nominal noise covariances are known. For the case where the statistic information of the nominal noise is absent, a modified expectation–maximization algorithm is introduced and integrated into the proposed robust smoother for simultaneously estimating the unknown states, system parameters, and the noise covariances. The performance of the proposed methods is illustrated by solving the aerodynamic coefficient identification problem with the real flight data set. The simulation results reveal that the proposed approaches outperform several existing solutions when outliers occur in multiple components of correlated measurements.

MHCVision: estimation of global and local false discovery rate for MHC class I peptide binding prediction.

MotivationMHC-peptide binding prediction has been widely used for understanding the immune response of individuals or populations, each carrying different MHC molecules as well as for the development of immunotherapeutics. The results from MHC-peptide binding prediction tools are mostly reported as a predicted binding affinity (IC50) and the percentile rank score, and global thresholds e.g. IC50 value < 500 nM or percentile rank < 2% are generally recommended for distinguishing binding peptides from non-binding peptides. However, it is difficult to evaluate statistically the probability of an individual peptide binding prediction to be true or false solely considering predicted scores. Therefore, statistics describing the overall global false discovery rate (FDR) and local FDR, also called posterior error probability (PEP) are required to give statistical context to the natively produced scores.ResultWe have developed an algorithm and code implementation, called MHCVision, for estimation of FDR and PEP values for the predicted results of MHC-peptide binding prediction from the NetMHCpan tool. MHCVision performs parameter estimation using a modified expectation maximization framework for a two-component beta mixture model, representing the distribution of true and false scores of the predicted dataset. We can then estimate the PEP of an individual peptide’s predicted score, and conversely the probability that it is true. We demonstrate that the use of global FDR and PEP estimation can provide a better trade-off between sensitivity and precision over using currently recommended thresholds from tools.Availability and implementation https://github.com/PGB-LIV/MHCVision.Supplementary information Supplementary data are available at Bioinformatics online.

A HYPER-SURFACE-BASED MODELING AND CORRECTION OF BIAS FIELD IN MR IMAGES

Dealing with the different artifacts in medical images is necessary to perform several tasks, in particular segmentation and registration. We introduce in this paper a novel method for bias field correction in Magnetic Resonance Images (MRI). Using the segmentation results obtained by a modified Expectation Maximization clustering, the bias field is fitted as an hyper-surface in a 4D hyper-space. Then, it is corrected based on the fact that voxels belonging to the same tissue should have the same intensity in the whole image. So, after a quick and coarse unsupervised voxel labeling by clustering by parts is performed, the bias field is computed at the voxels that have been reliably labeled as belonging to one of the tissues of interest. For the less reliably labeled voxels the bias field is interpolated using an hyper-surface, estimated by a 4D lagrangian interpolation. In order to evaluate the efficiency of the proposed method, we compare the MRI segmen- tation results with and without bias field correction, and we use also the coefficient of variations within the MRI volume.Segmentation results, and the coefficient of variations results were significantly enhanced after bias field correction by the proposed method. Furthermore, the proposed method is considered fast compared to other methods, that are often time consuming du to their iterative nature.

A deep convolutional neural network based computer aided diagnosis system for the prediction of Alzheimer's disease in MRI images

In the recent past, biomedical domain has become popular due to digital image processing of accurate and efficient diagnosis of clinical patients using Computer-Aided Diagnosis (CAD). Appropriate and punctual disease identification and treatment arrangement directs to enhance superiority of life and improved life hope in Alzheimer Disease (AD) patients. The cutting-edge approaches that believe multimodal analysis have been shown to be efficient and accurate are improved compared with manual analysis. Many tools have been introduced for detection of Alzheimer but still it is a financially high costly diagnosis system gives detection of disease with low accuracy and efficient due to performance of Magnetic Resonance Imaging (MRI) scanning devices. A novel methodology is proposed in this research as CAD process using various algorithms for predicting AD. The MRI images from scanning device are a highly noisy image due to thermal activities of hardware involved in scanning device. The image restoration technique is applied using 2D Adaptive Bilateral Filter (2D-ABF) algorithm. The quality of image in terms of brightness and contrast are improved using image enhancement techniques based on Adaptive Histogram Adjustment (AHA) algorithm. The Region of Interest of Alzheimer disease is segmented using Adaptive Mean Shift Modified Expectation Maximization (AMS-MEM) algorithm. The various features are calculated using second order 2-Dimensional Gray Level Co-Occurrence Matrix (2D-GLCM). Based on selection of features, the Deep Learning (DL) approach is used to classify the disease images and its stages. The Deep Convolutional Neural Network (DCNN) is the classification technique implemented to classify disease for proper diagnostic decision making. The experimental results prove that the proposed methodology provides better accuracy and efficiency than existing system.

Learning-Based Iterative Interference Cancellation for Cognitive Internet of Things

This paper is concerned with a machine learning approach to cancel the interference for cognitive Internet of Things (C-IoT) in the concurrent spectrum access (CSA) model, where the C-IoT system is noncooperative and has very limited knowledge on the interference. Our transceiver design uses an iterative processing structure, which consists of a linear estimator, a demodulation-and-decoding module, and a clustering module. In the clustering module, we employ modified expectation–maximization (EM)-based algorithms to estimate the interference under the knowledge of the modulation constraint (MC) of the interference. We show that this modified EM algorithm-based receiver outperforms the original EM-based receiver, since the former is able to generate a more accurate clustering result by reducing the dimension of the parameter space. We further improve the performance of the iterative receiver by introducing the extrinsic information technique, with the resulting algorithm referred to as the extrinsic modulation constrained EM (Ext-MC-EM) algorithm. We show that the Ext-MC-EM algorithm-based receiver considerably outperforms the counterpart iterative receivers, including the MC-EM algorithm.

Investigation of the optimal number of clusters by the adaptive EM algorithm

This paper considers the investigation of the optimal number of clusters for datasets that are modeled as the Gaussian mixture. For that purpose, the adaptive method that is based on the modified Expectation Maximization (EM) algorithm is developed. The modification is conducted within the hidden variable of the standard EM algorithm. Assuming that data are multivariate normally distributed, where each component of the Gaussian mixture corresponds to one cluster, the modification is provided by utilizing the fact that the Mahalanobis distance of samples follows a Chi-square distribution. Besides, the quantity measure is constructed in order to determine number of clusters. The proposed method is presented in several numerical examples.

A Note on Parameters Estimation for Nonlinear Wiener Processes With Measurement Errors

The nonlinear Wiener process has been widely used as a model for the degradation process. This note concerns parameters estimation of nonlinear Wiener processes with measurement errors (WPME) by the maximum likelihood estimation method. Firstly, we prove a rule that the estimated results based on the sample likelihood function developed through observations at each point are equal to the results from the first differences of the observations. This rule indicates that for reducing computation complexity the first differences of the observations may develop the sample likelihood function. Then we present a simple method to calculate the determinant and the inverse matrix of the covariance matrix of the WPME. This simple method could avoid the overflow error when calculating the determinant of the covariance matrix and the case that the inverse matrix is close to be singular, which could result in wrong estimation results. Secondly, we highlight the unit-specific assumption, which has a significant impact on parameters estimation but has been neglected in many papers. Then, we propose a modified expectation maximization algorithm for parameters estimation with random effects. Finally, to demonstrate the application and superiority of the proposed method, we provide a numerical example and a case study with comparison to several representative methods in the literature.

A new two-layer mixture of factor analyzers with joint factor loading model for the classification of small dataset problems

Abstract Dimensionality Reduction (DR) is a fundamental topic of pattern classification and machine learning. For classification tasks, DR is typically employed as a pre-processing step, succeeded by an independent classifier training stage. However, such independent operation of the two stages often limits the final classification performance notably, as the generated subspace may not be maximally beneficial or appropriate to the learning task at hand. This problem is further accentuated for high-dimensional data classification in situations of the limited number of samples. To address this problem, we develop a novel joint learning model for classification, referred to as two-layer mixture of factor analyzers with joint factor loading (2L-MJFA). Specifically, the model adopts a special two-layer mixture or a mixture of mixtures structure, where each component represents each specific class as a mixture of factor analyzers (MFA). Importantly, all the involved factor analyzers are intentionally designed so that they share the same loading matrix. This, apart from operating as the DR matrix, largely reduces the parameters and makes the proposed algorithm very suitable to small dataset situations. Additionally, we propose a modified expectation maximization algorithm to train the proposed model. A series of simulation experiments demonstrate that what we propose significantly outperforms other state-of-the-art algorithms on various benchmark datasets. Finally, since factor analyzers are closely linked with Auto-encoder networks, the proposed idea could be of particular utility to the community of neural networks.

Statistical Inference and Applications of Mixture of Varying Coefficient Models

AbstractIn this paper, we consider a new mixture of varying coefficient models, in which each mixture component follows a varying coefficient model and the mixing proportions and dispersion parameters are also allowed to be unknown smooth functions. We systematically study the identifiability, estimation and inference for the new mixture model. The proposed new mixture model is rather general, encompassing many mixture models as its special cases such as mixtures of linear regression models, mixtures of generalized linear models, mixtures of partially linear models and mixtures of generalized additive models, some of which are new mixture models by themselves and have not been investigated before. The new mixture of varying coefficient model is shown to be identifiable under mild conditions. We develop a local likelihood procedure and a modified expectation–maximization algorithm for the estimation of the unknown non‐parametric functions. Asymptotic normality is established for the proposed estimator. A generalized likelihood ratio test is further developed for testing whether some of the unknown functions are constants. We derive the asymptotic distribution of the proposed generalized likelihood ratio test statistics and prove that the Wilks phenomenon holds. The proposed methodology is illustrated by Monte Carlo simulations and an analysis of a CO2‐GDP data set.

Lung cancer segmentation and diagnosis of lung cancer staging using MEM (modified expectation maximization) algorithm and artificial neural network fuzzy inference system (ANFIS)

Medical image processing plays a major role in advancing treatment planning, identification of disease and guiding for surgery. Lung cancer is one among the dangerous diseases that leads to death of most human beings due to uncontrolled growth in the cell. This research area is finding more importance among researchers is that because the available methods for lung cancer detection are very painful. So the automatic detection of lung nodule of computer tomography (CT) is finding more importance among researchers. Though there are a number of methods available for segmentation of lung nodules each method has its drawbacks because of the divergence that exists in lung nodule. The proposed method is the detection of lung nodules using MEM algorithm and classification is based on Artificial Neural Network Fuzzy Inference System (ANFIS) experimental results illustrate in an improvement of accuracy and reduction of false positive results. A comparative analysis is performed with FCM and FLICM algorithms to show the superior nature of MEM.

Robust subspace clustering via penalized mixture of Gaussians

Abstract Many problems in computer vision and pattern recognition can be posed as learning low-dimensional subspace structures from high-dimensional data. Subspace clustering represents a commonly utilized subspace learning strategy. The existing subspace clustering models mainly adopt a deterministic loss function to describe a certain noise type between an observed data matrix and its self-expressed form. However, the noises embedded in practical high-dimensional data are generally non-Gaussian and have much more complex structures. To address this issue, this paper proposes a robust subspace clustering model by embedding the Mixture of Gaussians (MoG) noise modeling strategy into the low-rank representation (LRR) subspace clustering model. The proposed MoG-LRR model is capitalized on its adapting to a wider range of noise distributions beyond current methods due to the universal approximation capability of MoG. Additionally, a penalized likelihood method is encoded into this model to facilitate selecting the number of mixture components automatically. A modified Expectation Maximization (EM) algorithm is also designed to infer the parameters involved in the proposed PMoG-LRR model. The superiority of our method is demonstrated by extensive experiments on face clustering and motion segmentation datasets.

Measurement and improvement across immunization information systems: Paving pathways for pharmacy data exchange

Many problems in computer vision and pattern recognition can be posed as learning low-dimensional subspace structures from high-dimensional data. Subspace clustering represents a commonly utilized subspace learning strategy. The existing subspace clustering models mainly adopt a deterministic loss function to describe a certain noise type between an observed data matrix and its self-expressed form. However, the noises embedded in practical high-dimensional data are generally non-Gaussian and have much more complex structures. To address this issue, this paper proposes a robust subspace clustering model by embedding the Mixture of Gaussians (MoG) noise modeling strategy into the low-rank representation (LRR) subspace clustering model. The proposed MoG-LRR model is capitalized on its adapting to a wider range of noise distributions beyond current methods due to the universal approximation capability of MoG. Additionally, a penalized likelihood method is encoded into this model to facilitate selecting the number of mixture components automatically. A modified Expectation Maximization (EM) algorithm is also designed to infer the parameters involved in the proposed PMoG-LRR model. The superiority of our method is demonstrated by extensive experiments on face clustering and motion segmentation datasets.