Unobserved Data Research Articles

Bayesian hidden Markov model for natural history of colorectal cancer: Handling misclassified observations, varying observation schemes and unobserved data

Bayesian hidden Markov model for natural history of colorectal cancer: Handling misclassified observations, varying observation schemes and unobserved data

Clustering Functional Data With Measurement Errors: A Simulation-Based Approach.

Clustering analysis of functional data, which comprises observations that evolve continuously over time or space, has gained increasing attention across various scientific disciplines. Practical applications often involve functional data that are contaminated with measurement errors arising from imprecise instruments, sampling errors, or other sources. These errors can significantly distort the inherent data structure, resulting in erroneous clustering outcomes. In this article, we propose a simulation-based approach designed to mitigate the impact of measurement errors. Our proposed method estimates the distribution of functional measurement errors through repeated measurements. Subsequently, the clustering algorithm is applied to simulated data generated from the conditional distribution of the unobserved true functional data given the observed contaminated functional data, accounting for the adjustments made to rectify measurement errors. We illustrate through simulations show that the proposed method has improved numerical performance than the naive methods that neglect such errors. Our proposed method was applied to a childhood obesity study, giving more reliable clustering results.

The Measurement of the Gross Domestic Product affected by the shadow economy

The article presents a method for balancing Gross Domestic Product (GDP) when the measurements of its components are distorted by the existence of the shadow economy. Our proposal to measure GDP is based on a multiple ultrastructural model (MUM), where the explanatory variables are subject to error. We show that the expected value of GDP can be divided into two parts: the first part concerns data related to registered activities and the second part concerns unobserved data which may be partly related to the shadow economy. The empirical analysis is based on the annual data for individual voivodeships in Poland for the years 2000–2019. The data are obtained from the Local Data Bank of Statistics Poland. Two approaches to measuring GDP are considered: from the expenditure side and from the production side. The results show that the unobservable part of the variables necessary to balance GDP on the production side does not exceed 1% of GDP, and on the expenditure side, it mostly reaches about 3% of GDP.

Data Completion-Guided Unified Graph Learning for Incomplete Multi-View Clustering

Due to its heterogeneous property, multi-view data has been widely concerned over single-view data for performance improvement. Unfortunately, some instances may be with partially available information because of some uncontrollable factors, for which the incomplete multi-view clustering (IMVC) problem is raised. IMVC aims to partition unlabeled incomplete multi-view data into their clusters by exploiting the heterogeneity of multi-view data and overcoming the difficulty of data loss. However, most existing IMVC methods like BSV, MIC, OMVC, and IVC tend to conduct basic completion processing on the input data, without taking advantage of the correlation between samples and information redundancy. To overcome the above issue, we propose one novel IMVC method named data completion-guided unified graph learning (DCUGL), which could complete the data of missing views and fuse multiple learned view-specific similarity matrices into one unified graph. Specifically, we first reduce the dimension of the input data to learn multiple view-specific similarity matrices. By stacking all view-specific similarity matrices, DCUGL constructs a third-order tensor with the low-rank constraint, such that sample correlation within and between views can be well explored. Finally, by dividing the original data into observed data and unobserved data, DCUGL can infer and complete the missing data according to the view-specific similarity matrices, and obtain a unified graph, which can be directly used for clustering. To solve the proposed model, we design an iterative algorithm, which is based on the alternating direction method of multipliers framework. The proposed model proves to be superior by benchmarking on six challenging datasets compared with state-of-the-art IMVC methods.

Transfer learning through physics-informed neural networks for bubble growth in superheated liquid domains

In this paper, a physics-informed neural network (PINN) technique is developed to study the heat and mass transfer for the process of vapour bubble growth in a superheated liquid domain and tested using three working fluids including water, R-134a and FC-72. The work represents a novel step in the development of PINNs for phase change scenarios where surface tension effects dominate, and acts as a necessary validation stage before PINN techniques can be applied to complex boiling analysis. Initially, a forward analysis was performed using water and R-134a as working fluids. For each of these investigations, the PINN algorithm was trained on 50 % of the available CFD data. The proposed algorithm was able to accurately infer velocity fields, particularly in the near-interfacial region. The resultant circulatory flow was found to maintain the desired circular shape of the growing bubbles. As a result, when predicting the evolution of a water vapour bubble, the developed PINN algorithm produced a reduction in peak error by 0.87 % compared to CFD reference data, and 3.42 % reduction in peak error for prediction of the evolution of the R-134a vapour bubble. To test and optimise the transfer learning capabilities of the developed methodology, the evolution of an FC-72 vapour bubble in superheated FC-72 was predicted without supplying supporting observational data. For this scenario, the PINN algorithm produced a peak error within 1.3 % of the unobserved CFD reference data. The proposed approach confirms the robustness of PINN methodologies as a method of solving phase-change problems where surface tension plays a pivotal, promising to expedite parametric studies in practice. This study represents a pioneering effort in the development of PINNs for phase change by applying the current algorithm to investigate bubble growth within superheated liquid domains, serving as a basis for the application of PINNs for boiling problems and as a benchmark for inverse training strategy.

Marginal inference for hierarchical generalized linear mixed models with patterned covariance matrices using the Laplace approximation

AbstractWe develop hierarchical models and methods in a fully parametric approach to generalized linear mixed models for any patterned covariance matrix. The Laplace approximation is used to marginally estimate covariance parameters by integrating over all fixed and latent random effects. The Laplace approximation relies on Newton–Raphson updates, which also leads to predictions for the latent random effects. We develop methodology for complete marginal inference, from estimating covariance parameters and fixed effects to making predictions for unobserved data. The marginal likelihood is developed for six distributions that are often used for binary, count, and positive continuous data, and our framework is easily extended to other distributions. We compare our methods to fully Bayesian methods, automatic differentiation, and integrated nested Laplace approximations (INLA) for bias, mean‐squared (prediction) error, and interval coverage, and all methods yield very similar results. However, our methods are much faster than Bayesian methods, and more general than INLA. Examples with binary and proportional data, count data, and positive‐continuous data are used to illustrate all six distributions with a variety of patterned covariance structures that include spatial models (both geostatistical and areal models), time series models, and mixtures with typical random intercepts based on grouping.

Next location prediction using heterogeneous graph-based fusion network with physical and social awareness

Location prediction based on social media information is highly valuable in human mobility research and has multiple real-life applications. However, existing research methods often ignore social influences, largely ignoring implicit information regarding interactions between users and geographical locations. Additionally, they generally employ single modeling structures, which restricts the effective integration of complex spatiotemporal characteristics and factors influencing user mobility. In this context, we propose a novel network with physical and social awareness that expresses both physical and social influences of user mobility from a global perspective based on a heterogeneous graph constructed using users and spatial locations as nodes and relationships between them as edges. This graph enables the model to leverage information from connected nodes and edges to infer missing or unobserved data. The model predicts future locations of users by effectively integrating the temporal and spatial features of user trajectory series. The proposed model is validated using three social media datasets. The experimental results demonstrate that the proposed method outperforms the state-of-the-art baseline models. This indicates the importance of considering complex interactions between users and locations, as well as the various influences of physical and social spaces.

How to treat mixed behavior segments in supervised machine learning of behavioural modes from inertial measurement data

The application of supervised machine learning methods to identify behavioural modes from inertial measurements of bio-loggers has become a standard tool in behavioural ecology. Several design choices can affect the accuracy of identifying the behavioural modes. One such choice is the inclusion or exclusion of segments consisting of more than a single behaviour (mixed segments) in the machine learning model training data. Currently, the common practice is to ignore such segments during model training. In this paper we tested the hypothesis that including mixed segments in model training will improve accuracy, as the model would perform better in identifying them in the test data. We test this hypothesis using a series of data simulations on four datasets of accelerometer data coupled with behaviour observations, obtained from four study species (Damaraland mole-rats, meerkats, olive baboons, polar bears). Results show that when a substantial proportion of the test data are mixed behaviour segments (above ~ 10%), including mixed segments in machine learning model training improves the accuracy of classification. These results were consistent across the four study species, and robust to changes in segment length, sample size, and degree of mixture within the mixed segments. However, we also find that in some cases (particularly in baboons) models trained with mixed segments show reduced accuracy in classifying test data containing only single behaviour (pure) segments, compared to models trained without mixed segments. Based on these results, we recommend that when the classification model is expected to deal with a substantial proportion of mixed behaviour segments (> 10%), it is beneficial to include them in model training, otherwise, it is unnecessary but also not harmful. The exception is when there is a basis to assume that the training data contains a higher rate of mixed segments than the actual (unobserved) data to be classified—such a situation may occur particularly when training data are collected in captivity and used to classify data from the wild. In this case, excess inclusion of mixed segments in training data should probably be avoided.

Modelling the developmental path in the acquisition of /l/-final irregular plurals by Portuguese children

ABSTRACT Previous empirical research has shown that Portuguese children aged 4;0 to 6;0 are sensitive to the quality of stem-final vowels when acquiring the irregular plural forms of /l/-final words (acquisition order: plurals of /al, ɛl, ɔl, ul/ > plurals of /il/). This study presents a formal account of this emergent pattern. We first construct an initial-state grammar that arguably instantiates Portuguese children’s grammatical knowledge at the onset of morpho-phonological acquisition. We then simulate morpho-phonological learning using two constraint-based models (Stochastic Optimality Theory and Noisy Harmonic Grammar) and their associated Gradual Learning Algorithm. The results of our learning simulations corroborate the experimental data, showing that the plural form of /il/-final words takes longer to master than that of other /l/-final words. In addition to replicating the empirical results, the learning simulation reveals two important implications. First, the raw frequency of individual forms cannot account for the attested patterns, suggesting that the input frequency is mediated by some principles of grammatical learning. Second, Harmonic Grammar outperforms Optimality Theory in simulating real acquisition data. Through the additive constraint interaction, Harmonic-Grammar learning is more restrictive, avoiding the generation of unobserved data in the course of morpho-phonological acquisition.

Abstract 901: BATCHIE: An active learning platform for scalable combination drug screens

Abstract Large-scale combination drug screens are largely considered intractable due to the immense number of possible combinations. Existing approaches use ad hoc fixed experimental designs then train machine learning models to impute novel combinations. We introduce BATCHIE, an orthogonal approach that adaptively conducts experiments in batches. BATCHIE uses information theory and probabilistic modeling to design each batch to be maximally informative based on the results of previous experiments. BATCHIE is fully modular, allowing any Bayesian probabilistic model to be used and any study constraints to be incorporated while maintaining optimality guarantees. Results: In retrospective simulations on public combination screens, BATCHIE saved 10s of thousands to 100s of thousands of experiments relative to non-adaptive baselines. We conducted a prospective study focusing on pediatric sarcomas with BATCHIE. Our study covered 16 cell lines spanning Ewing sarcoma (EWS), osteosarcoma, rhabdomyosarcoma, as well as non-sarcoma cancers and non-cancer lines. We used a drug library of 206 drugs at two doses. After 15 rounds of BATCHIE-driven data collection, we observed 54K unique combinations, covering 4% of the experimental landscape. On unobserved validation data, the BATCHIE model predictions were highly accurate (Pearson’s rho=0.91, p<10-30) and detected the rare (0.004%) combinations with significant synergy (AUC of ROC=0.85, p<10-5). We further investigated 10 combinations that BATCHIE predicted to have high therapeutic index (TI) for a broad selection of EWS lines, meaning a large differential effect between the predicted viabilities of the control lines and the target lines. We found that the TI scores for the top hits were significantly larger than the rest of the screen (p<10−50), with the median top hit TI score lying in the 98th percentile of observed TI scores. We further validated 6 of the top hits in an ex vivo study on 2 patient-derived EWS samples, again finding significantly large TI values (p<10-13), with the median ex vivo TI score lying in the 96th percentile of observed TI scores. The top hits also exhibited biologically plausible rationales including combining PARP inhibitors with topoisomerase 1 inhibitors and alkylating agents, despite our model utilizing no prior knowledge on the molecular targets of our drug library. Combining PARP inhibitors with topoisomerase 1 inhibitors comprise 3 of the 6 currently open phase II clinical combination trials for EWS. Citation Format: Christopher Tosh, Mauricio Tec, Jessica White, Jeffrey F. Quinn, Glorymar Ibanez Sanchez, Paul Calder, Andrew L. Kung, Filemon S. Dela Cruz, Wesley Tansey. BATCHIE: An active learning platform for scalable combination drug screens [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 901.

Optimizing time prediction and error classification in early melanoma detection using a hybrid RCNN-LSTM model.

Skin cancer is a terrifying disorder that affects all individuals. Due to the significant increase in the rate of melanoma skin cancer, early detection of skin cancer is now more critical than ever before. Malignant melanoma is one of the most serious forms of skin cancer, and it is caused by abnormal melanocyte cell growth. In recent years, skin cancer predictive categorization has become more accurate and predictive due to multiple deep learning algorithms. Malignant melanoma is diagnosed using the Recurrent Convolution Neural Network-Long Short-Term Memory (RCNN-LSTM), which is one of the deep learning classification approaches. Using the International Skin Image Collection and the RCNN-LSTM, the data are categorized and analyzed to gain a better understanding of skin cancer. The method begins with data preprocessing, which prepares the dataset for classification. Additionally, the RCNN is employed to extract the features that are vital to the prediction process. The LSTM is accountable for the final step, classification. There are further factors to examine, such as the precision of 94.60%, the sensitivity of 95.67%, and the F1-score of 95.13%. Other benefits of the suggested study include shorter prediction durations of 95.314, 122.530, and 131.205 s and lower model loss of 0.25%, 0.19%, and 0.15% for input sizes 10, 15, and 20, respectively. Three datasets had a reduced categorization error of 5.11% and an accuracy of 95.42%. In comparison to previous approaches, the work discussed here produces superior outcomes. RESEARCH HIGHLIGHTS: Recurrent convolutional neural network (RCNN) deep learning approach for optimizing time prediction and error classification in early melanoma detection. It extracts a high number of specific features from the skin disease image, making the classification process easier and more accurate. To reduce classification errors in accurately detecting melanoma, context dependency is considered in this work. By accounting for context dependency, the deprivation state is avoided, preventing performance degradation in the model. To minimize melanoma detection model loss, a skin disease image augmentation or regularization process is performed in this work. This strategy improves the accuracy of the model when applied to fresh, previously unobserved data.

On the Impact of CDL and TDL Augmentation for RF Fingerprinting Under Impaired Channels

Cyber-physical systems have recently been used in several areas (such as connected and autonomous vehicles) due to their high maneuverability. On the other hand, they are susceptible to cyber-attacks. Radio frequency (RF) fingerprinting emerges as a promising approach. This work aims to analyze the impact of decoupling tapped delay line and clustered delay line (TDL+CDL) augmentation-driven deep learning (DL) on transmitter-specific fingerprints to discriminate malicious users from legitimate ones. This work also considers 5G-only-CDL, WiFi-only-TDL augmentation approaches. RF fingerprinting models are sensitive to changing channels and environmental conditions. For this reason, they should be considered during the deployment of a DL model. Data acquisition can be another option. Nonetheless, gathering samples under various conditions for a train set formation may be quite hard. Consequently, data acquisition may not be feasible. This work uses a dataset that includes 5G, 4G, and WiFi samples, and it empowers a CDL+TDL based augmentation technique in order to boost the learning performance of the DL model. Numerical results show that CDL+TDL, 5G-only-CDL, and WiFi-only-TDL augmentation approaches achieve 87.59%, 81.63%, 79.21% accuracy on unobserved data while TDL/CDL augmentation technique and no augmentation approach result in 77.81% and 74.84% accuracy on unobserved data, respectively.

Assessing the multivariate distributional accuracy of common imputation methods

Imputation methods are popular tools that allow for a wide range of subsequent analyses on complete data sets. However, in order for these analyses to be trustworthy, it is important that the imputation procedure reflects the true distribution of the unobserved data sufficiently well. This raises the question how well different imputation methods can reproduce multivariate correlations, associations or even the entire multivariate distribution. The paper gives first answers to this question by means of an extensive comparative simulation study. In particular, we evaluate the multivariate distributional accuracy for six state-of-the art imputation algorithms with respect to different measures and give practical recommendations.

PECJ: Stream Window Join on Disorder Data Streams with Proactive Error Compensation

Stream Window Join (SWJ), a vital operation in stream analytics, struggles with achieving a balance between accuracy and latency due to out-of-order data arrivals. Existing methods predominantly rely on adaptive buffering, but often fall short in performance, thereby constraining practical applications. We introduce PECJ, a solution that proactively incorporates unobserved data to enhance accuracy while reducing latency, thus requiring robust predictive modeling of stream oscillation. At the heart of PECJ lies a mathematical formulation of the posterior distribution approximation (PDA) problem using variational inference (VI). This approach circumvents error propagation while meeting the low-latency demands of SWJ. We detail the implementation of PECJ, striking a balance between complexity and generality, and discuss both analytical and learning-based approaches. Experimental evaluations reveal PECJ's superior performance. The successful integration of PECJ into a multi-threaded SWJ benchmark testbed further establishes its practical value, demonstrating promising advancements in enhancing data stream processing capabilities amidst out-of-order data.

A modified generative adversarial networks with Yolov5 for automated forest health diagnosis from aerial imagery and Tabu search algorithm.

Our environment has been significantly impacted by climate change. According to previous research, insect catastrophes induced by global climate change killed many trees, inevitably contributing to forest fires. The condition of the forest is an essential indicator of forest fires. Analysis of aerial images of a forest can detect deceased and living trees at an early stage. Automated forest health diagnostics are crucial for monitoring and preserving forest ecosystem health. Combining Modified Generative Adversarial Networks (MGANs) and YOLOv5 (You Only Look Once version 5) is presented in this paper as a novel method for assessing forest health using aerial images. We also employ the Tabu Search Algorithm (TSA) to enhance the process of identifying and categorizing unhealthy forest areas. The proposed model provides synthetic data to supplement the limited labeled dataset, thereby resolving the frequent issue of data scarcity in forest health diagnosis tasks. This improvement enhances the model's ability to generalize to previously unobserved data, thereby increasing the overall precision and robustness of the forest health evaluation. In addition, YOLOv5 integration enables real-time object identification, enabling the model to recognize and pinpoint numerous tree species and potential health issues with exceptional speed and accuracy. The efficient architecture of YOLOv5 enables it to be deployed on devices with limited resources, enabling forest-monitoring applications on-site. We use the TSA to enhance the identification of unhealthy forest areas. The TSA method effectively investigates the search space, ensuring the model converges to a near-optimal solution, improving disease detection precision and decreasing false positives. We evaluated our MGAN-YOLOv5 method using a large dataset of aerial images of diverse forest habitats. The experimental results demonstrated impressive performance in diagnosing forest health automatically, achieving a detection precision of 98.66%, recall of 99.99%, F1 score of 97.77%, accuracy of 99.99%, response time of 3.543ms and computational time of 5.987ms. Significantly, our method outperforms all the compared target detection methods showcasing a minimum improvement of 2% in mAP.

SetRank: A Setwise Bayesian Approach for Collaborative Ranking in Recommender System

The recent development of recommender systems has a focus on collaborative ranking, which provides users with a sorted list rather than rating prediction. The sorted item lists can more directly reflect the preferences for users and usually perform better than rating prediction in practice. While considerable efforts have been made in this direction, the well-known pairwise and listwise approaches have still been limited by various challenges. Specifically, for the pairwise approaches, the assumption of independent pairwise preference is not always held in practice. Also, the listwise approaches cannot efficiently accommodate “ties” and unobserved data due to the precondition of the entire list permutation. To this end, in this article, we propose a novel setwise Bayesian approach for collaborative ranking, namely, SetRank, to inherently accommodate the characteristics of user feedback in recommender systems. SetRank aims to maximize the posterior probability of novel setwise preference structures and three implementations for SetRank are presented. We also theoretically prove that the bound of excess risk in SetRank can be proportional to \(\sqrt {M/N}\) , where M and N are the numbers of items and users, respectively. Finally, extensive experiments on four real-world datasets clearly validate the superiority of SetRank compared with various state-of-the-art baselines.

Dominance of posterior predictive densities over plug-in densities for order statistics in exponential distributions

Many researchers have proposed numerous Bayesian predictive densities for Type-II censored data that is generated by ordered observations. However, their evaluations of predictive densities were insufficient because the Bayesian predictive density includes prior parameters, thus we suffer from the selection of the prior parameters. In this study, we consider two types of predictive densities, posterior predictive and plug-in, for observations from an exponential distribution of Type-II censored data. We discuss a suitable predictive density using the risk with the Kullback–Leibler loss function. In our setting, we consider a Gamma prior, which is a conjugate prior for mathematical tractability. We prove that the posterior predictive density with an improper Gamma prior provides the dominance of the posterior predictive density over the plug-in densities without depending on the selection of an unknown parameter in our setting. Finally, we show that the posterior predictive density outperforms the plug-in densities in terms of coverage probabilities for unobserved data by censoring in a simulation study.

Estimation of Dormant Cell Population in Cancer Patients: A New Approach

The branching processes form a configuration for modeling tumor cells. Faced with unobserved data on dormant cells, inference based on the branching process is not easy to achieve. In large populations, we construct a new framework for estimating dormant cells and tumor dormancy rates. This inference uses of control theory is based on deterministic process statistics approximating branching process in large populations. Precisely, we use an auxiliary system called an observer whose solutions tend exponentially towards those of the limit deterministic model. This observer uses only available measurable data on tumor cells and provides estimates of the number of dormant cells. In addition, the constructed observer does not use the parameter of the generally unknown tumor dormancy rate. We also derive a method to estimate it using the estimated states. We apply this estimation method using simulated data from the branching process.

Regression of binary network data with exchangeable latent errors

AbstractUndirected, binary network data consist of indicators of symmetric relations between pairs of actors. Regression models of such data allow for the estimation of effects of exogenous covariates on the network and for prediction of unobserved data. Ideally, estimators of the regression parameters should account for the inherent dependencies among relations in the network that involve the same actor. To account for such dependencies, researchers have developed a host of latent variable network models; however, estimation of many latent variable network models is computationally onerous and which model is best to base inference upon may not be clear. We propose the probit exchangeable (PX) model for undirected binary network data that is based on an assumption of exchangeability, which is common to many of the latent variable network models in the literature. The PX model can represent the first two moments of any exchangeable network model. We leverage the EM algorithm to obtain an approximate maximum likelihood estimator of the PX model that is extremely computationally efficient. Using simulation studies, we demonstrate the improvement in estimation of regression coefficients of the proposed model over existing latent variable network models. In an analysis of purchases of politically aligned books, we demonstrate political polarization in purchase behavior and show that the proposed estimator significantly reduces runtime relative to estimators of latent variable network models, while maintaining predictive performance.

Being Bayesian about learning Gaussian Bayesian networks from incomplete data

We propose a Bayesian model averaging (BMA) approach for inferring the structure of Gaussian Bayesian networks (BNs) from incomplete data, i.e. from data with missing values. Our method builds on the ‘Bayesian metric for Gaussian networks having score equivalence’ (BGe score) and we make the assumption that the unobserved data points are ‘missing completely at random’. We present a Markov Chain Monte Carlo sampling algorithm that allows for simultaneously sampling directed acyclic graphs (DAGs) as well as the values of the unobserved data points. We empirically cross-compare the network reconstruction accuracy of the new BMA approach with two non-Bayesian approaches for dealing with incomplete BN data, namely the classical structural Expectation Maximisation (EM) approach and the more recently proposed node average likelihood (NAL) method. For the empirical evaluation we use synthetic data from a benchmark Gaussian BN and real wet-lab protein phosphorylation data from the RAF signalling pathway.