Practical Datasets Research Articles

A New INAR(1) Model for ℤ-Valued Time Series Using the Relative Binomial Thinning Operator

Abstract A new first-order integer-valued autoregressive process (INAR(1)) with extended Poisson innovations is introduced based on a signed version of the thinning operator, called relative binomial thinning operator, which can be considered as an extension of standard binomial thinning operator introduced by Steutel, F.W. and van Harn, K. (1979. Discrete analogues of self-decomposability and stability. Ann. Probab. 7: 893–899). It is appropriate for modeling Z $\mathbb{Z}$ -valued time series and either positive or negative correlations. Some properties of the process are established. Conditional least squares, Yule–Walker and conditional maximum likelihood methods are considered for the parameter estimation of the model. Moreover, simulation experiments are carried out to attest to the performance of the estimation methods. The applicability of the proposed model is investigated through a practical data set of the Saudi stock market.

A novel shape classification method using 1‐D convolutional neural networks

AbstractMost of the shape classification methods are based on a single closed contour. However, practical shapes always have complex contours, for example, a combination of multiple open contours. How to accurately identify complex shapes is an unsolved problem. In this research, a novel method is proposed to classify complex shapes. The proposed method firstly encodes a complex shape to an angle code and a sparsity code, then input these codes to a 1‐D CNN for extracting features and classification. Experiments on two datasets show this novel method is superior in terms of classification accuracy. These two datasets are practical shape dataset collected by this paper on internet and MPEG‐7 CE‐1 Part B. The proposed method achieves higher classification accuracy than compared methods. In order to show the performance of the proposed method on each class, the accuracy on each class is analyzed. Ablation experiment is conducted to show the contribution of each module in the network. The result shows that each module is meaningful in the network, because without any module the accuracy drops.

Evaluation of an artificial intelligence–based algorithm for automated localization of craniofacial landmarks

ObjectivesDue to advancing digitalisation, it is of interest to develop standardised and reproducible fully automated analysis methods of cranial structures in order to reduce the workload in diagnosis and treatment planning and to generate objectifiable data. The aim of this study was to train and evaluate an algorithm based on deep learning methods for fully automated detection of craniofacial landmarks in cone-beam computed tomography (CBCT) in terms of accuracy, speed, and reproducibility.Materials and methodsA total of 931 CBCTs were used to train the algorithm. To test the algorithm, 35 landmarks were located manually by three experts and automatically by the algorithm in 114 CBCTs. The time and distance between the measured values and the ground truth previously determined by an orthodontist were analyzed. Intraindividual variations in manual localization of landmarks were determined using 50 CBCTs analyzed twice.ResultsThe results showed no statistically significant difference between the two measurement methods. Overall, with a mean error of 2.73 mm, the AI was 2.12% better and 95% faster than the experts. In the area of bilateral cranial structures, the AI was able to achieve better results than the experts on average.ConclusionThe achieved accuracy of automatic landmark detection was in a clinically acceptable range, is comparable in precision to manual landmark determination, and requires less time.Clinical relevanceFurther enlargement of the database and continued development and optimization of the algorithm may lead to ubiquitous fully automated localization and analysis of CBCT datasets in future routine clinical practice.

MTrajPlanner: A Multiple-Trajectory Planning Algorithm for Autonomous Underwater Vehicles

Trajectory planning is a crucial task in designing the navigation systems of automatic underwater vehicles (AUVs). Due to the complexity of underwater environments, decision makers may hope to obtain multiple alternative trajectories in order to select the best. This paper focuses on the multiple-trajectory planning (MTP) problem, which is a new topic in this field. First, we establish a comprehensive MTP model for AUVs, by taking into account the complex underwater environments, the efficiency of each trajectory, and the diversity among different trajectories, simultaneously. Then, to solve the MTP, we develop an ant colony-based trajectory optimizer, which is characterized by a niching strategy, a decayed alarm pheromone measure, and a diversified heuristic measure. The niching strategy assists in identifying and maintaining a diverse set of high-quality solutions. The use of decayed alarm pheromone and diversified heuristic further improves the search effectiveness and efficiency of the algorithm. Experimental results on practical datasets show that our proposed algorithm not only provides multiple AUV trajectories for a flexible choice, but it also outperforms the state-of-the-art algorithms in terms of the single trajectory efficiency.

Motif-guided Heterogeneous Graph Deep Generation.

The complex systems in the real-world are commonly associated with multiple types of objects and relations, and heterogeneous graphs are ubiquitous data structures that can inherently represent multi-modal interactions between objects. Generating high-quality heterogeneous graphs allows us to understand the implicit distribution of heterogeneous graphs and provides benchmarks for downstream heterogeneous representation learning tasks. Existing works are limited to either merely generating the graph topology with neglecting local semantic information or only generating the graph without preserving the higher-order structural information and the global heterogeneous distribution in generated graphs. To this end, we formulate a general, end-to-end framework - HGEN for generating novel heterogeneous graphs with a newly proposed heterogeneous walk generator. On top of HGEN, we further develop a network motif generator to better characterize the higher-order structural distribution. A novel heterogeneous graph assembler is further developed to adaptively assemble novel heterogeneous graphs from the generated heterogeneous walks and motifs in a stratified manner. The extended model is proven to preserve the local semantic and heterogeneous global distribution of observed graphs with the theoretical guarantee. Lastly, comprehensive experiments on both synthetic and real-world practical datasets demonstrate the power and efficiency of the proposed method.

Network-aware RF-energy harvesting for designing energy efficient IoT networks

Energy Harvesting plays a crucial role in improving the network lifetime of an IoT application. Radio Frequency (RF) energy is one of the most prominent ambient energy sources available that can be used for harvesting energy. Moreover, a number of dedicated RF-energy transmitters is required for sufficient energy transfer to the power deficit IoT nodes. However, the appropriate positioning of these RF-energy transmitters in the network is a challenging issue. In this context, a network-aware RF-energy transmitter positioning scheme is proposed in this work. This RF-energy transmitter deployment scheme considers different network performance parameters such as energy-hole information, node-connectivity information, and data routing information for RF-energy transmitter placement. Apart from that, the amount of RF-energy harvested is constrained to the RF-energy carried by the propagating RF-waves. Moreover, the wireless medium uncertainties result in severe attenuation of the propagating RF-waves. In this regard, a robust hybrid RF-wave propagation model is also proposed. The proposed model amalgamates different wave propagation mechanisms for signal strength prediction in the propagating RF-waves. This helps in improving the modelling accuracy of the RF-waves. The proposed model also predicts the amount of residual RF-energy available in the propagating RF-wave for harvesting. This model is also validated against the practical datasets generated from experiments. The proposed transmitter positioning scheme, along with the hybrid RF-wave propagation model, overcome the energy hole issue in the IoT networks and improves energy efficiency. Extensive simulations are conducted under various network scenarios to evaluate the energy harvesting performance of the proposed methods. It is observed that the proposed network-aware RF-energy harvesting scheme improves the network performance in terms of energy harvesting, network coverage, and the lifetime of IoT network.

TS-LSSVM: Triple sparse least squares support vector machine for residual oxygen concentration detection of encapsulated pharmaceutical vials

Accurate measurement of residual oxygen concentration in encapsulated pharmaceutical vials is adequate to ensure the quality of inner sterile preparations. However, the critical characteristic signal is feeble and covered by enormous environmental interference in the actual production. Inspired by structural sparse learning, we propose a novel prediction model in this paper, triple sparse least squares support vector machine (TS-LSSVM), in which the production priors are deeply excavated, and the feature, sample, and structure sparsity are realized simultaneously by redefining the objective function. In addition, the selection of support vectors can be adjusted adaptively according to the time-varying environmental noise, so as to ensure the reliability of long-term operation. First, the time–frequency components containing the prior knowledge are extracted based on the synchro squeezing wavelet transform (SSWT). Then, a triple sparse learning strategy is designed, which can accurately eliminate redundant wavelet coefficients and adaptively select training samples. Finally, a strict and fast optimization solution is proposed under the alternating direction method of multipliers (ADMM) framework. Experimental results on public and practical datasets prove the superiority of TS-LSSVM.

The Harris Extended Bilal Distribution with Applications in Hydrology and Quality Control

In this research work, a new three-parameter lifetime distribution is introduced and studied. It is called the Harris extended Bilal distribution due to its construction from a mixture of the famous Bilal and Harris distributions, resulting from a branching process. The basic properties, such as the moment generating function, moments, quantile function, and Rényi entropy, are discussed. We show that the hazard rate function has ideal features for modeling increasing, upside-down bathtub, and roller-coaster data sets. In a second part, the Harris extended Bilal model is investigated from a statistical viewpoint. The maximum likelihood estimation is used to estimate the parameters, and a simulation study is carried out. The flexibility of the proposed model in a hydrological data analysis scenario is demonstrated using two practical data sets and compared with important competing models. After that, we establish an acceptance sampling plan that takes advantage of all of the features of the Harris extended Bilal model. The operating characteristic values, the minimum sample size that corresponds to the maximum possible defects, and the minimum ratios of lifetime associated with the producer’s risk are discussed.

Optimal Quantum Dataset for Learning a Unitary Transformation

Unitary transformations formulate the time evolution of quantum states. How to learn a unitary transformation efficiently is a fundamental problem in quantum machine learning. The most natural and leading strategy is to train a quantum machine learning model based on a quantum dataset. Although the presence of more training data results in better models, using too much data reduces the efficiency of training. In this work, we solve the problem on the minimum size of sufficient quantum datasets for learning a unitary transformation exactly, which reveals the power and limitation of quantum data. First, we prove that the minimum size of a dataset with pure states is $2^n$ for learning an $n$-qubit unitary transformation. To fully explore the capability of quantum data, we introduce a practical quantum dataset consisting of $n+1$ elementary tensor product states that are sufficient for exact training. The main idea is to simplify the structure utilizing decoupling, which leads to an exponential improvement in the size of the datasets with pure states. Furthermore, we show that the size of the quantum dataset with mixed states can be reduced to a constant, which yields an optimal quantum dataset for learning a unitary. We showcase the applications of our results in oracle compiling and Hamiltonian simulation. Notably, to accurately simulate a 3-qubit one-dimensional nearest-neighbor Heisenberg model, our circuit only uses $96$ elementary quantum gates, which is significantly less than $4080$ gates in the circuit constructed by the Trotter-Suzuki product formula.

Open-Set Fault Diagnosis via Supervised Contrastive Learning With Negative Out-of-Distribution Data Augmentation

Fault diagnosis in an open world refers to the diagnosis tasks that need to cope with previously unknown faults in the online stage. It faces a great challenge yet to be addressed—that is, the online data of unknown faults may be classified as normal samples with a high probability. In this article, we develop an effective solution for this challenge by using supervised contrastive learning to learn a discriminative and compact embedding for the known normal situation and fault situations. Specifically, in addition to contrasting a given sample with other instances as is the case in conventional contrastive learning methods, our training scheme contrasts the normal samples with negative augmentations of themselves. The negative out-of-distribution data is generated by the Soft Brownian Offset sampling method to simulate the previously unknown faults. Computational experiments are conducted on the Tennessee Eastman Process benchmark dataset and a practical plasma etching process dataset. The proposed method achieves significant improvement compared with four existing methods under three open-set fault diagnosis circumstances, i.e., balanced open-set fault diagnosis, imbalanced fault diagnosis, and few-shot fault diagnosis. This demonstrates its great potentials in real world fault diagnosis applications.

MSC-CSMC: A multi-objective semi-supervised clustering algorithm based on constraints selection and multi-source constraints for gene expression data.

Many clustering techniques have been proposed to group genes based on gene expression data. Among these methods, semi-supervised clustering techniques aim to improve clustering performance by incorporating supervisory information in the form of pairwise constraints. However, noisy constraints inevitably exist in the constraint set obtained on the practical unlabeled dataset, which degenerates the performance of semi-supervised clustering. Moreover, multiple information sources are not integrated into multi-source constraints to improve clustering quality. To this end, the research proposes a new multi-objective semi-supervised clustering algorithm based on constraints selection and multi-source constraints (MSC-CSMC) for unlabeled gene expression data. The proposed method first uses the gene expression data and the gene ontology (GO) that describes gene annotation information to form multi-source constraints. Then, the multi-source constraints are applied to the clustering by improving the constraint violation penalty weight in the semi-supervised clustering objective function. Furthermore, the constraints selection and cluster prototypes are put into the multi-objective evolutionary framework by adopting a mixed chromosome encoding strategy, which can select pairwise constraints suitable for clustering tasks through synergistic optimization to reduce the negative influence of noisy constraints. The proposed MSC-CSMC algorithm is testified using five benchmark gene expression datasets, and the results show that the proposed algorithm achieves superior performance.

Interaction with Industrial Digital Twin Using Neuro-Symbolic Reasoning.

Digital twins have revolutionized manufacturing and maintenance, allowing us to interact with virtual yet realistic representations of the physical world in simulations to identify potential problems or opportunities for improvement. However, traditional digital twins do not have the ability to communicate with humans using natural language, which limits their potential usefulness. Although conventional natural language processing methods have proven to be effective in solving certain tasks, neuro-symbolic AI offers a new approach that leads to more robust and versatile solutions. In this paper, we propose neuro-symbolic reasoning (NSR)-a fundamental method for interacting with 3D digital twins using natural language. The method understands user requests and contexts to manipulate 3D components of digital twins and is able to read maintenance manuals and implement installations and removal procedures autonomously. A practical neuro-symbolic dataset of machine-understandable manuals, 3D models, and user queries is collected to train the neuro-symbolic reasoning interaction mechanism. The evaluation demonstrates that NSR can execute user commands accurately, achieving 96.2% accuracy on test data. The proposed method has industrial importance since it provides the technology to perform maintenance procedures, request information from manuals, and serve as a tool to interact with complex virtual machinery using natural language.

A constrained proper orthogonal decomposition model for upscaling permeability

AbstractReservoir modeling and simulation are vital components of modern reservoir management processes. Despite the advances in computing power and the advent of smart technologies, the implementation of model‐based operational/control strategies has been limited by the inherent complexity of reservoir models. Thus, reduce order models that not only reduce the cost of the implementation but also provide geological consistent prediction are essential. This article introduces reduced‐order models based on the proper orthogonal decomposition (POD) coupled with linear interpolation for upscaling. First, using POD‐based models, low rank approximate (LRA) are obtained by projecting the high dimensional permeability dataset to a low dimensional subspace spanned by its trajectories to decorrelate the dataset. Next, the LRA is integrated into the interpolation algorithm to generate upscaled values. This technique is highly scalable since low‐rank approximations are achieved by the variation in the number of modes used for reconstruction. To test the validity and reliability of the model, we show its application to the practical dataset from SPE10 benchmark case2. From statistics of the error analysis, the classical POD algorithm seems to be more preferred for LRA; however, since non‐negativity of the permeability data set is a priority, the constrained POD (non‐negative POD) algorithm described in this article is more appropriate. Finally, we compared the POD‐based models to a traditional industry‐standard upscaling technique (e.g., arithmetic mean) to highlight our model benefits/performance. Results show that the POD‐based models, particularly the non‐negative POD model, produce considerably less error than the arithmetic mean model in the upscaling process.

Agreement of acute serious events recorded across datasets using linked Australian general practice, hospital, emergency department and death data: implications for research and surveillance.

Understanding the level of recording of acute serious events in general practice electronic health records (EHRs) is critical for making decisions about the suitability of general practice datasets to address research questions and requirements for linking general practice EHRs with other datasets. To examine data source agreement of five serious acute events (myocardial infarction, stroke, venous thromboembolism (VTE), pancreatitis and suicide) recorded in general practice EHRs compared with hospital, emergency department (ED) and mortality data. Data from 61 general practices routinely contributing data to the MedicineInsight database was linked with New South Wales administrative hospital, ED and mortality data. The study population comprised patients with at least three clinical encounters at participating general practices between 2019 and 2020 and at least one record in hospital, ED or mortality data between 2010 and 2020. Agreement was assessed between MedicineInsight diagnostic algorithms for the five events of interest and coded diagnoses in the administrative data. Dates of concordant events were compared. The study included 274,420 general practice patients with at least one record in the administrative data between 2010 and 2020. Across the five acute events, specificity and NPV were excellent (>98%) but sensitivity (13%-51%) and PPV (30%-75%) were low. Sensitivity and PPV were highest for VTE (50.9%) and acute pancreatitis (75.2%), respectively. The majority (roughly 70-80%) of true positive cases were recorded in the EHR within 30 days of administrative records. Large proportions of events identified from administrative data were not detected by diagnostic algorithms applied to general practice EHRs within the specific time period. EHR data extraction and study design only partly explain the low sensitivities/PPVs. Our findings support the use of Australian general practice EHRs linked to hospital, ED and mortality data for robust research on the selected serious acute conditions.

User Behavior Simulation for Search Result Re-ranking

Result ranking is one of the major concerns for Web search technologies. Most existing methodologies rank search results in descending order of relevance. To model the interactions among search results, reinforcement learning (RL algorithms have been widely adopted for ranking tasks. However, the online training of RL methods is time and resource consuming at scale. As an alternative, learning ranking policies in the simulation environment is much more feasible and efficient. In this article, we propose two different simulation environments for the offline training of the RL ranking agent: the Context-aware Click Simulator (CCS) and the Fine-grained User Behavior Simulator with GAN (UserGAN). Based on the simulation environment, we also design a User Behavior Simulation for Reinforcement Learning (UBS4RL) re-ranking framework, which consists of three modules: a feature extractor for heterogeneous search results, a user simulator for collecting simulated user feedback, and a ranking agent for generation of optimized result lists. Extensive experiments on both simulated and practical Web search datasets show that (1) the proposed user simulators can capture and simulate fine-grained user behavior patterns by training on large-scale search logs, (2) the temporal information of user searching process is a strong signal for ranking evaluation, and (3) learning ranking policies from the simulation environment can effectively improve the search ranking performance.

Energy Theft Detection Model Based on VAE-GAN for Imbalanced Dataset

Energy theft causes a lot of economic losses every year. In the practical environment of energy theft detection, it is required to solve imbalanced data problem where normal user data are significantly larger than energy theft data. In this paper, a variational autoencoder-generative adversarial network (VAE-GAN)-based energy theft-detection model is proposed to overcome the imbalanced data problem. In the proposed model, the VAE-GAN generates synthetic energy theft data with the features of real energy theft data for augmenting the energy theft dataset. The obtained balanced dataset is applied to train a detector which is designed as one-dimensional convolutional neural network. The proposed model is simulated on the practical dataset for comparing with various generative models to evaluate their performance. From simulation results, it is confirmed that the proposed model outperforms the other existing models. Additionally, it is shown that the proposed model is also very useful in the environments of extreme data imbalance for a wide variety of applications by analyzing the performance of detector according to the balance rate.

A novel total nitrogen prediction method based on recurrent neural networks utilizing cross-coupling attention and selective attention

Considering the critical role the total nitrogen (TN) plays in the stable operation of wastewater treatment plants (WWTPs), it is necessary to predict its future variation accurately. However, limited by current detection techniques, only a few kinds of water quality parameters (WQPs) with finite historical data can be obtained in the WWTPs. The small sample size of WQPs tremendously hinders the precision of TN prediction. In this study, a novel cross-coupling attention recurrent neural network (CCA-RNN) is proposed to overcome this problem. First, cross-coupling attention (CCA) is designed to enable the topological structures in the historical input data to be directly and effectively extracted with fewer training samples. Then, selective attention is introduced to dynamically select the useful topological relationships and the corresponding variables sent by the high-speed channel. Compared with the other state-of-the-art methods, CCA-RNN achieves the best performance on both the public and practical wastewater datasets, proving its superiority and great potential to be deployed in similar problems.

MALS-Net: A Multi-Head Attention-Based LSTM Sequence-to-Sequence Network for Socio-Temporal Interaction Modelling and Trajectory Prediction

Predicting the trajectories of surrounding vehicles is an essential task in autonomous driving, especially in a highway setting, where minor deviations in motion can cause serious road accidents. The future trajectory prediction is often not only based on historical trajectories but also on a representation of the interaction between neighbouring vehicles. Current state-of-the-art methods have extensively utilized RNNs, CNNs and GNNs to model this interaction and predict future trajectories, relying on a very popular dataset known as NGSIM, which, however, has been criticized for being noisy and prone to overfitting issues. Moreover, transformers, which gained popularity from their benchmark performance in various NLP tasks, have hardly been explored in this problem, presumably due to the accumulative errors in their autoregressive decoding nature of time-series forecasting. Therefore, we propose MALS-Net, a Multi-Head Attention-based LSTM Sequence-to-Sequence model that makes use of the transformer's mechanism without suffering from accumulative errors by utilizing an attention-based LSTM encoder-decoder architecture. The proposed model was then evaluated in BLVD, a more practical dataset without the overfitting issue of NGSIM. Compared to other relevant approaches, our model exhibits state-of-the-art performance for both short and long-term prediction.

Provably accurate and scalable linear classifiers in hyperbolic spaces

Many high-dimensional practical data sets have hierarchical structures induced by graphs or time series. Such data sets are hard to process in Euclidean spaces, and one often seeks low-dimensional embeddings in other space forms to perform the required learning tasks. For hierarchical data, the space of choice is a hyperbolic space because it guarantees low-distortion embeddings for tree-like structures. The geometry of hyperbolic spaces has properties not encountered in Euclidean spaces that pose challenges when trying to rigorously analyze algorithmic solutions. We propose a unified framework for learning scalable and simple hyperbolic linear classifiers with provable performance guarantees. The gist of our approach is to focus on Poincaré ball models and formulate the classification problems using tangent space formalisms. Our results include a new hyperbolic perceptron algorithm as well as an efficient and highly accurate convex optimization setup for hyperbolic support vector machine classifiers. Furthermore, we adapt our approach to accommodate second-order perceptrons, where data are preprocessed based on second-order information (correlation) to accelerate convergence, and strategic perceptrons, where potentially manipulated data arrive in an online manner and decisions are made sequentially. The excellent performance of the Poincaré second-order and strategic perceptrons shows that the proposed framework can be extended to general machine learning problems in hyperbolic spaces. Our experimental results, pertaining to synthetic, single-cell RNA-seq expression measurements, CIFAR10, Fashion-MNIST and mini-ImageNet, establish that all algorithms provably converge and have complexity comparable to those of their Euclidean counterparts.

EMOSTUDENT: A DATASET FOR COMPLEX STUDENT BEHAVIOUR EVALUATION

This article discusses artificial intelligence based automated online proctoring systems. The practical implementation of the task of analysing students’ behaviour in the process of working with these systems and the datasets used to solve this task were considered. A general model for processing data on human activity in the process of online learning has been created, which is aimed at analysing and describing the activity and emotional state of the student. For this purpose, various features that affect the assessment of student behaviour during independent work on test tasks or exams were identified. Based on the analysis of existing datasets and the problems of modern implementations of automated online proctoring systems, a classification of features used specifically for analysing video sequences in the context of solving the problem of analysing student behaviour was made. In accordance with the developed requirements, a dataset was proposed. The data source for the developed dataset was the YouTube platform: videos with a Creative Commons licence were used. Amazon SageMaker platform was utilized to organise the process of data labelling and dataset formation. The generated dataset was added to the Kaggle and Hugging Face platforms. This allows us to utilized the work among other scientists and software developers and to test the developed dataset in practice in training of various implemented models of artificial neural networks.