Data Analysis Tasks Research Articles

Perspectives and Challenges of AI Techniques in the Field of Social Sciences and Communication

<div><p class="4">In the past decade, the methods and technologies of artificial intelligence (AI) have made great progress. In many cases, they have become part of the usual landscape of solving new or old problems in different fields of human knowledge. In this progress, there are several aspects, especially three aspects: the availability and universality of data in many fields of human activities; a deeper understanding of the mathematics of the basic control algorithm; and the availability and capability of hardware and computing which allows a wide range and a large number of data experiments. Considering these aspects, the key challenge for each problem and application area is to understand how to use these technologies, to what extent they may reach, and what constraints need to be overcome in order to obtain beneficial results (in terms of production cost, value, etc.). This challenge includes identifying data sources and their integration and recovery requirements, the necessity and cost of acquiring or constructing tag data sets, volume data required for measurement, verifying its feasibility, technical method of data analysis task and its consistency with the final application goal, and social and communication sciences are no exception. The knowledge in these fields is related to artificial intelligence, but they do have particularities that define the most appropriate type of artificial intelligence technology and method (i.e. natural language processing). The successful use of AI technology in these disciplines involves not only technical knowledge, but also the establishment of a viable application environment, including the availability of data, the appropriate complexity of tasks to be performed, and verification procedures with experts in the field. This paper introduces the methodology of generating artificial intelligence model, summarizes the artificial intelligence methods and services most likely to be used in social and communication sciences, and finally gives some application examples to illustrate the practical and technical considerations in this regard.</p></div>

Dowker complex based machine learning (DCML) models for protein-ligand binding affinity prediction.

With the great advancements in experimental data, computational power and learning algorithms, artificial intelligence (AI) based drug design has begun to gain momentum recently. AI-based drug design has great promise to revolutionize pharmaceutical industries by significantly reducing the time and cost in drug discovery processes. However, a major issue remains for all AI-based learning model that is efficient molecular representations. Here we propose Dowker complex (DC) based molecular interaction representations and Riemann Zeta function based molecular featurization, for the first time. Molecular interactions between proteins and ligands (or others) are modeled as Dowker complexes. A multiscale representation is generated by using a filtration process, during which a series of DCs are generated at different scales. Combinatorial (Hodge) Laplacian matrices are constructed from these DCs, and the Riemann zeta functions from their spectral information can be used as molecular descriptors. To validate our models, we consider protein-ligand binding affinity prediction. Our DC-based machine learning (DCML) models, in particular, DC-based gradient boosting tree (DC-GBT), are tested on three most-commonly used datasets, i.e., including PDBbind-2007, PDBbind-2013 and PDBbind-2016, and extensively compared with other existing state-of-the-art models. It has been found that our DC-based descriptors can achieve the state-of-the-art results and have better performance than all machine learning models with traditional molecular descriptors. Our Dowker complex based machine learning models can be used in other tasks in AI-based drug design and molecular data analysis.

Hybrid Parallel Stochastic Configuration Networks for Industrial Data Analytics

As a class of randomized learner model, stochastic configuration networks (SCNs) have been successfully applied in a few data analytics tasks. Given the industrial big data modeling tasks, however, the original SCNs potentially lead to excessive training time. To this end, this article extends SCNs to a hybrid parallel version, termed hybrid parallel SCNs (HPSCNs). In the hybrid parallel learning algorithm, two SCNs are synchronously constructed. The difference between the two of them is that one employs a point-incremental algorithm, and another one adopts a block-incremental algorithm. Moreover, a data parallel method is established based on a dynamic block strategy to accelerate the establishment of candidate node pool for each SCN. Additionally, this article proposes an adaptive hyperparameter adjustment method, which allows the hyperparameters in the supervisory mechanism to be automatically adjusted along with the learning process. Comparative experiments are first conducted through four large-scale benchmark datasets, followed by the fully discussion on the algorithm parameters. Finally, a practical industrial application case is made, where a grinding particle size soft sensor is developed based on HPSCN, showing the effectiveness of the proposed algorithm.

SuRVoS 2: Accelerating Annotation and Segmentation for Large Volumetric Bioimage Workflows Across Modalities and Scales.

As sample preparation and imaging techniques have expanded and improved to include a variety of options for larger sized and numbers of samples, the bottleneck in volumetric imaging is now data analysis. Annotation and segmentation are both common, yet difficult, data analysis tasks which are required to bring meaning to the volumetric data. The SuRVoS application has been updated and redesigned to provide access to both manual and machine learning-based segmentation and annotation techniques, including support for crowd sourced data. Combining adjacent, similar voxels (supervoxels) provides a mechanism for speeding up segmentation both in the painting of annotation and by training a segmentation model on a small amount of annotation. The support for layers allows multiple datasets to be viewed and annotated together which, for example, enables the use of correlative data (e.g. crowd-sourced annotations or secondary imaging techniques) to guide segmentation. The ability to work with larger data on high-performance servers with GPUs has been added through a client-server architecture and the Pytorch-based image processing and segmentation server is flexible and extensible, and allows the implementation of deep learning-based segmentation modules. The client side has been built around Napari allowing integration of SuRVoS into an ecosystem for open-source image analysis while the server side has been built with cloud computing and extensibility through plugins in mind. Together these improvements to SuRVoS provide a platform for accelerating the annotation and segmentation of volumetric and correlative imaging data across modalities and scales.

A machine learning toolkit for CRISM image analysis

Hyperspectral images collected by remote sensing have played a significant role in the discovery of aqueous alteration minerals, which in turn have important implications for our understanding of the changing habitability on Mars. Traditional spectral analyzes based on summary parameters have been helpful in converting hyperspectral cubes into readily visualizable three channel maps highlighting high-level mineral composition of the Martian terrain. These maps have been used as a starting point in the search for specific mineral phases in images. Although the amount of labor needed to verify the presence of a mineral phase in an image is quite limited for phases that emerge with high abundance, manual processing becomes laborious when the task involves determining the spatial extent of detected phases or identifying small outcrops of secondary phases that appear in only a few pixels within an image. Thanks to extensive use of remote sensing data and rover expeditions, significant domain knowledge has accumulated over the years about mineral composition of several regions of interest on Mars, which allow us to collect reliable labeled data required to train machine learning algorithms. In this study we demonstrate the utility of machine learning in two essential tasks for hyperspectral data analysis: nonlinear noise removal and mineral classification. We develop a simple yet effective hierarchical Bayesian model for estimating distributions of spectral patterns and extensively validate this model for mineral classification on several test images. Our results demonstrate that machine learning can be highly effective in exposing tiny outcrops of specific phases in orbital data that are not uncovered by traditional spectral analysis. We package implemented scripts, documentation illustrating use cases, and pixel-scale training data collected from dozens of well-characterized images into a new toolkit. We hope that this new toolkit will provide advanced and effective processing tools and improve community’s ability to map compositional units in remote sensing data quickly, accurately, and at scale.

Interactive and Reproducible Workflows for Exploring and Modeling RNA-seq Data with pcaExplorer, Ideal, and GeneTonic.

The generation and interpretation of results from transcriptome profiling experiments via RNA sequencing (RNA-seq) can be a complex task. While raw data quality control, alignment, and quantification can be streamlined via efficient algorithms that can deliver the preprocessed expression matrix, a common bottleneck in the analysis of such large datasets is the subsequent in-depth, iterative processes of data exploration, statistical testing, visualization, and interpretation. Specific tools for these workflow steps are available but require a level of technical expertise which might be prohibitive for life and clinical scientists, who are left with essential pieces of information distributed among different tabular and list formats. Our protocols are centered on the joint use of our Bioconductor packages (pcaExplorer, ideal, GeneTonic) for interactive and reproducible workflows. All our packages provide an interactive and accessible experience via Shiny web applications, while still documenting the steps performed with RMarkdown as a framework to guarantee the reproducibility of the analyses, reducing the overall time to generate insights from the data at hand. These protocols guide readers through the essential steps of Exploratory Data Analysis, statistical testing, and functional enrichment analyses, followed by integration and contextualization of results. In our packages, the core elements are linked together in interactive widgets that make drill-down tasks efficient by viewing the data at a level of increased detail. Thanks to their interoperability with essential classes and gold-standard pipelines implemented in the open-source Bioconductor project and community, these protocols will permit complex tasks in RNA-seq data analysis, combining interactivity and reproducibility for following modern best scientific practices and helping to streamline the discovery process for transcriptome data. © 2022 The Authors. Current Protocols published by Wiley Periodicals LLC. Basic Protocol 1: Exploratory Data Analysis with pcaExplorer Basic Protocol 2: Differential Expression Analysis with ideal Basic Protocol 3: Interpretation of RNA-seq results with GeneTonic Support Protocol: Downloading and installing pcaExplorer, ideal, and GeneTonic Alternate Protocol: Using functions from pcaExplorer, ideal, and GeneTonic in custom analyses.

Efficient Multimedia Frame-Skipping Architecture Using Deep Learning in Vehicular Networks

With the development of 5G networks, vehicle-to-vehicle communication is helping to make travel safe. However, vehicle type detection in the multimedia feed remains a problem. This helps reduce processing time and enable more dynamic connectivity between different vehicles. The development of object classes requires more robust computer vision models and algorithms. However, the main difficulty still lies in the image quality, which depends on the lighting conditions, viewing angle, and physical structure of the vehicles. This research mainly focuses on the development and deployment of a deep learning-based system for traffic congestion analysis. The model uses multiple video feeds and vehicle information to detect, classify, and count vehicles in the live traffic feed. The model is trained with a deep learning approach to align the video image and detect the object in top-down multimedia. The dynamic skipping method helps to process a long video feed and accurately compares the video image with the viewer. The standard query for the vehicle can help in recognizing and creating the models in real-time traffic situations. The proposed model is suitable for many applications that require a specific area for monitoring real-time data analysis and multimedia routine tasks.

A critical re-evaluation of neural methods for entity alignment

Neural methods have become the de-facto choice for the vast majority of data analysis tasks, and entity alignment (EA) is no exception. Not surprisingly, more than 50 different neural EA methods have been published since 2017. However, surprisingly, an analysis of the differences between neural and non-neural EA methods has been lacking. We bridge this gap by performing an in-depth comparison among five carefully chosen representative state-of-the-art methods from the pre-neural and neural era. We unravel, and consequently mitigate, the inherent deficiencies in the experimental setup utilized for evaluating neural EA methods. To ensure fairness in evaluation, we homogenize the entity matching modules of neural and non-neural methods. Additionally, for the first time, we draw a parallel between EA and record linkage (RL) by empirically showcasing the ability of RL methods to perform EA. Our results indicate that Paris, the state-of-the-art non-neural method, statistically significantly outperforms all the representative state-of-the-art neural methods in terms of both efficacy and efficiency across a wide variety of dataset types and scenarios, and is second only to BERT-INT for a specific scenario of cross-lingual EA. Our findings shed light on the potential problems resulting from an impulsive application of neural methods as a panacea for all data analytics tasks. Overall, our work results in two overarching conclusions: (1) Paris should be used as a baseline in every follow-up work on EA, and (2) neural methods need to be positioned better to showcase their true potential, for which we provide multiple recommendations.

Software for processing and analysis of experimental data in researching of gas sensors

Representation of structural elements, appearance and use of software for processing experimental data on the kinetics of the response of resistive gas sensors based on the KWW stretched exponential function model algorithm. The developed software allows the use of input data in the form of text files or Excel spreadsheets, which include special applications of measuring equipment. When using applications, the application of the concept of using various software packages and environments is most effective for each task of data processing and analysis. The software allows the subsequent loading of data to measure the kinetic sensitivity of the gas sensor with the fixation of the parameters for measuring the level of its measurement (temperature sensor, partial pressure of the gas increase). The processing results can be rep-resented by the dependences of the parameters of the KWW function on the operating pa-rameter. The application provides services such as data entry using Excel spreadsheet programs; cal-culations in the mathematical package "Mathcad"; test examples of the use of computational algorithms and the transmission of reference information texts from circulation and the Inter-net.

Pattern Recognition Characteristics and Neural Mechanism of Basketball Players’ Dribbling Tactics Based on Artificial Intelligence and Deep Learning

There are many factors that affect a player’s overall basketball ability, and different factors will have different effects. The effect is mainly manifested in the difference of offensive and defensive data of basketball players in basketball games. In the basketball field, the formulation of existing training plans mainly relies on the manual observation and personal experience of coaches. This method is inevitably subjective. The pattern recognition of dribbling tactics is one of the important factors. A proper dribbling tactic can make the team achieve better results. In order to discover different dribbling characteristics, reanalyze the connotation and manifestation of basketball speed and strive to analyze the factors that affect basketball speed reasonably and accurately. The deep learning algorithm simulates the thinking process of the human brain neurons through the computer method and then realizes the function of the computer to automatically learn the data characteristics and complete the complex data analysis task. We use artificial intelligence and deep learning to simulate various dribbling tactics of players and find out the rules to improve players’ abilities. The results of the study prove that developing a suitable dribbling tactical model for basketball players can increase their competitive ability by more than 10%, reduce the damage to players, and prolong their careers. Generally speaking, athletes’ injuries can be reduced by more than 15%. This shows that the pattern recognition characteristics and neural mechanisms of dribbling tactics are extremely important to basketball players.

Phenonizer: A Fine-Grained Phenotypic Named Entity Recognizer for Chinese Clinical Texts.

Biomedical named entity recognition (BioNER) from clinical texts is a fundamental task for clinical data analysis due to the availability of large volume of electronic medical record data, which are mostly in free text format, in real-world clinical settings. Clinical text data incorporates significant phenotypic medical entities (e.g., symptoms, diseases, and laboratory indexes), which could be used for profiling the clinical characteristics of patients in specific disease conditions (e.g., Coronavirus Disease 2019 (COVID-19)). However, general BioNER approaches mostly rely on coarse-grained annotations of phenotypic entities in benchmark text dataset. Owing to the numerous negation expressions of phenotypic entities (e.g., “no fever,” “no cough,” and “no hypertension”) in clinical texts, this could not feed the subsequent data analysis process with well-prepared structured clinical data. In this paper, we developed Human-machine Cooperative Phenotypic Spectrum Annotation System (http://www.tcmai.org/login, HCPSAS) and constructed a fine-grained Chinese clinical corpus. Thereafter, we proposed a phenotypic named entity recognizer: Phenonizer, which utilized BERT to capture character-level global contextual representation, extracted local contextual features combined with bidirectional long short-term memory, and finally obtained the optimal label sequences through conditional random field. The results on COVID-19 dataset show that Phenonizer outperforms those methods based on Word2Vec with an F1-score of 0.896. By comparing character embeddings from different data, it is found that character embeddings trained by clinical corpora can improve F-score by 0.0103. In addition, we evaluated Phenonizer on two kinds of granular datasets and proved that fine-grained dataset can boost methods' F1-score slightly by about 0.005. Furthermore, the fine-grained dataset enables methods to distinguish between negated symptoms and presented symptoms. Finally, we tested the generalization performance of Phenonizer, achieving a superior F1-score of 0.8389. In summary, together with fine-grained annotated benchmark dataset, Phenonizer proposes a feasible approach to effectively extract symptom information from Chinese clinical texts with acceptable performance.

A Wavelet-Based Independence Test for Functional Data With an Application to MEG Functional Connectivity

Measuring and testing the dependency between multiple random functions is often an important task in functional data analysis. In the literature, a model-based method relies on a model which is subject to the risk of model misspecification, while a model-free method only provides a correlation measure which is inadequate to test independence. In this paper, we adopt the Hilbert–Schmidt Independence Criterion (HSIC) to measure the dependency between two random functions. We develop a two-step procedure by first pre-smoothing each function based on its discrete and noisy measurements and then applying the HSIC to recovered functions. To ensure the compatibility between the two steps such that the effect of the pre-smoothing error on the subsequent HSIC is asymptotically negligible when the data are densely measured, we propose a new wavelet thresholding method for pre-smoothing and to use Besov-norm-induced kernels for HSIC. We also provide the corresponding asymptotic analysis. The superior numerical performance of the proposed method over existing ones is demonstrated in a simulation study. Moreover, in a magnetoencephalography (MEG) data application, the functional connectivity patterns identified by the proposed method are more anatomically interpretable than those by existing methods.

Liu regression after random forest for prediction and modeling in high dimension

AbstractIn the modern era, using advanced technology, we have access to data with many features, and therefore, feature engineering has become a vital task in data analysis. One of the challenges in model estimation is to combat multicollinearity in high‐dimensional data problems where the number of features ( ) exceeds the number of samples . We propose a novel, yet simple, strategy to estimate the regression parameters in a high‐dimensional regime in the presence of multicollinearity. The proposed approach enjoys the good properties of the random forest and the simple structure of a class of linear unified estimators. We give a fast and straightforward algorithm to estimate the regression coefficients when and multicollinearity exist. Numerical investigation reveals the superior performance of the method in test mean squared error. The technique is also applied to melting chemical data, where we conducted an estimation among 4885 features and discussed advantages.

ScGate: marker-based purification of cell types from heterogeneous single-cell RNA-seq datasets.

SummaryA common bioinformatics task in single-cell data analysis is to purify a cell type or cell population of interest from heterogeneous datasets. Here, we present scGate, an algorithm that automatizes marker-based purification of specific cell populations, without requiring training data or reference gene expression profiles. scGate purifies a cell population of interest using a set of markers organized in a hierarchical structure, akin to gating strategies employed in flow cytometry. scGate outperforms state-of-the-art single-cell classifiers and it can be applied to multiple modalities of single-cell data (e.g. RNA-seq, ATAC-seq, CITE-seq). scGate is implemented as an R package and integrated with the Seurat framework, providing an intuitive tool to isolate cell populations of interest from heterogeneous single-cell datasets.Availability and implementationscGate is available as an R package at https://github.com/carmonalab/scGate (https://doi.org/10.5281/zenodo.6202614). Several reproducible workflows describing the main functions and usage of the package on different single-cell modalities, as well as the code to reproduce the benchmark, can be found at https://github.com/carmonalab/scGate.demo (https://doi.org/10.5281/zenodo.6202585) and https://github.com/carmonalab/scGate.benchmark. Test data are available at https://doi.org/10.6084/m9.figshare.16826071.Supplementary information Supplementary data are available at Bioinformatics online.

Releasing Differentially Private Trajectories with Optimized Data Utility

The ubiquity of GPS-enabled devices has resulted in an abundance of data about individual trajectories. Releasing trajectories enables a range of data analysis tasks, such as urban planning, but it also poses a risk in compromising individual location privacy. To tackle this issue, a number of location privacy protection algorithms are proposed. However, existing works are primarily concerned with maintaining the trajectory data geographic utility and neglect the semantic utility. Thus, many data analysis tasks relying on utility, e.g., semantic annotation, suffer from poor performance. Furthermore, the released trajectories are vulnerable to location inference attacks and de-anonymization attacks due to insufficient privacy guarantee. In this paper, to design a location privacy protection algorithm for releasing an offline trajectory dataset to potentially untrusted analyzers, we propose a utility-optimized and differentially private trajectory synthesizer (UDPT) with two novel features. First, UDPT simultaneously preserves both geographical utility and semantic utility by solving a data utility optimization problem with a genetic algorithm. Second, UDPT provides a formal and provable guarantee against privacy attacks by synthesizing obfuscated trajectories in a differentially private manner. Extensive experimental evaluations on real-world datasets demonstrate UDPT’s outperformance against state-of-the-art works in terms of data utility and privacy.

A national assessment of legacy versus new generation Medicaid data.

To compare the performance of Medicaid legacy, Medicaid new generation, and Medicare claims on data analytic tasks. Medicaid Analytic eXtract (MAX) claims (legacy) of 100% beneficiaries in 2011 (all states except Idaho), 2012 (all states), 2013 (28 states), and 2014 (17 states); 2016 Transformed Medicaid Statistical Information System Analytic Files (TAF) claims (new generation) of 100% beneficiaries from all states; Medicare claims of 20% beneficiaries in 2011-2014, 2016. We focused on the chain of events that starts with an out-of-hospital medical emergency and ends with hospital death or survival to discharge. We developed six data quality indicators to assess ambulance variables; linkage between claims; external cause of injury code reporting; and death reporting on hospital discharge status codes. For the latter, we estimated injury severity and modeled its association with death in the Medicare population. We used the model to compare reported versus expected deaths by injury severity in the Medicaid population. Datasets were compared by state and fee-for-service versus managed care. Medicare and Medicaid beneficiaries with emergency ambulance transports. Medicare claims had high performance across indicators and states; MAX claims substantially underperformed on multiple indicators in most states. For example, most states reported external cause codes for over 90% of Medicare but less than 15% of Medicaid injury cases. Medicaid fee-for-service did not consistently perform better than Medicaid managed care. Compared with MAX, TAF claims performed significantly better on some indicators but continued to have poor external cause code reporting. Finally, MAX and TAF managed care records reported deaths at discharge in the range of expected deaths; however, fee-for-service claims might have underreported high-severity injury deaths. New generation Medicaid claims performed better than legacy claims on some indicators, but much more improvement is needed to allow high-quality policy analysis.

Near real-time streaming analysis of big fusion data

Experiments on fusion plasmas produce high-dimensional data time series with ever-increasing magnitude and velocity, but turn-around times for analysis of this data have not kept up. For example, many data analysis tasks are often performed in a manual, ad-hoc manner some time after an experiment. In this article, we introduce the Delta framework that facilitates near real-time streaming analysis of big and fast fusion data. By streaming measurement data from fusion experiments to a high-performance compute center, Delta allows computationally expensive data analysis tasks to be performed in between plasma pulses. This article describes the modular and expandable software architecture of Delta and presents performance benchmarks of individual components as well as of an example workflow. Focusing on a streaming analysis workflow where electron cyclotron emission imaging (ECEi) data is measured at KSTAR on the National Energy Research Scientific Computing Center’s (NERSC’s) supercomputer we routinely observe data transfer rates of about 4 Gigabit per second. In NERSC, a demanding turbulence analysis workflow effectively utilizes multiple nodes and graphical processing units and executes them in under 5 min. We further discuss how Delta uses modern database systems and container orchestration services to provide web-based real-time data visualization. For the case of ECEi data we demonstrate how data visualizations can be augmented with outputs from machine learning models. By providing session leaders and physics operators, results of higher-order data analysis using live visualizations may make more informed decisions on how to configure the machine for the next shot.

JEBIN: analyzing gene co-expressions across multiple datasets by joint network embedding.

The inference of gene co-expression associations is one of the fundamental tasks for large-scale transcriptomic data analysis. Due to the high dimensionality and high noises in transcriptomic data, it is difficult to infer stable gene co-expression associations from single dataset. Meta-analysis of multisource data can effectively tackle this problem. We proposed Joint Embedding of multiple BIpartite Networks (JEBIN) to learn the low-dimensional consensus representation for genes by integrating multiple expression datasets. JEBIN infers gene co-expression associations in a nonlinear and global similarity manner and can integrate datasets with different distributions in linear time complexity with the gene and total sample size. The effectiveness and scalability of JEBIN were verified by simulation experiments, and its superiority over the commonly used integration methods was proved by three indexes on real biological datasets. Then, JEBIN was applied to study the gene co-expression patterns of hepatocellular carcinoma (HCC) based on multiple expression datasets of HCC and adjacent normal tissues, and further on latest HCC single-cell RNA-seq data. Results show that gene co-expressions are highly different between bulk and single-cell datasets. Finally, many differentially co-expressed ligand-receptor pairs were discovered by comparing HCC with adjacent normal data, providing candidate HCC targets for abnormal cell-cell communications.

Fair train-test split in machine learning: Mitigating spatial autocorrelation for improved prediction accuracy

Machine learning supports prediction and inference in multivariate and complex datasets where observations are spatially related to one another. Frequently, these datasets depict spatial autocorrelation that violates the assumption of identically and independently distributed data. Overlooking this correlation result in over-optimistic models that fail to account for the geographical configuration of data. Furthermore, although different data split methods account for spatial autocorrelation, these methods are inflexible, and the parameter training and hyperparameter tuning of the machine learning model is set with a different prediction difficulty than the planned real-world use of the model. In other words, it is an unfair training-testing process. We present a novel method that considers spatial autocorrelation and planned real-world use of the spatial prediction model to design a fair train-test split. Demonstrations include two examples of the planned real-world use of the model using a realistic multivariate synthetic dataset and the analysis of 148 wells from an undisclosed Equinor play. First, the workflow applies the semivariogram model of the target to compute the simple kriging variance as a proxy of spatial estimation difficulty based on the spatial data configuration. Second, the workflow employs a modified rejection sampling to generate a test set with similar prediction difficulty as the planned real-world use of the model. Third, we compare 100 test sets' realizations to the model's planned real-world use, using probability distributions and two divergence metrics: the Jensen-Shannon distance and the mean squared error. The analysis ranks the spatial fair train-test split method as the only one to replicate the difficulty (i.e., kriging variance) compared to the validation set approach and spatial cross-validation. Moreover, the proposed method outperforms the validation set approach, yielding a minor mean percentage error when predicting a target feature in an undisclosed Equinor play using a random forest model. The resulting outputs are training and test sets ready for model fit and assessment with any machine learning algorithm. Thus, the proposed workflow offers spatial aware sets ready for predictive machine learning problems with similar estimation difficulty as the planned real-world use of the model and compatible with any spatial data analysis task.

Node Classification in Attributed Multiplex Networks Using Random Walk and Graph Convolutional Networks

Node classification, as a central task in the graph data analysis, has been studied extensively with network embedding technique for single-layer graph network. However, there are some obstacles when extending the single-layer network embedding technique to the attributed multiplex network. The classification of a given node in the attributed multiplex network must consider the network structure in different dimensions, as well as rich node attributes, and correlations among the different dimensions. Moreover, the distance node context information of a given node in each dimension will also affect the classification of the given node. In this study, a novel network embedding approach for the node classification of attributed multiplex networks using random walk and graph convolutional networks (AMRG) is proposed. A random walk network embedding technique was used to extract distant node information and the results are considered as pre-trained node features to be concatenated with the original node features inputted into the graph convolutional networks (GCNs) to learn node representations for each dimension. Besides, the consensus regularization is introduced to capture the similarities among different dimensions, and the learnable neural network parameters of GCNs for different dimensions are also constrained by the regularization mechanism to improve the correlations. As well as an attention mechanism is explored to infer the importance for a given node in different dimensions. Extensive experiments demonstrated that our proposed technique outperforms many competitive baselines on several real-world multiplex network datasets.