Original Data Points Research Articles

US-LIME: Increasing fidelity in LIME using uncertainty sampling on tabular data

LIME has gained significant attention as an explainable artificial intelligence algorithm that sheds light on how complex machine learning models make decisions within a specific locality. One of the challenges of LIME is its instability and infidelity in acquiring explanations in multiple runs. This study focuses on improving LIME's fidelity by presenting a new sampling strategy. The idea is to generate more focused data samples close to the decision boundary and simultaneously close to the original data point (the sample targeted to be explained). Then, these newly concentrated data are used to train a simple and interpretable linear model as an alternative to the original complex model. The approach leads to high-quality and local sample generation and thus improves the overall fidelity of the model while preserving the constancy of the explanations compared to competing methods. The superiority of the proposed method is shown through comprehensive experiments and comparing the results with LIME, LS-LIME, S-LIME, and BayLIME in terms of fidelity while maintaining stability. This method also performs better than BayLIME, SLIME, and LS-LIME algorithms in terms of execution time. In addition, tests related to the effect of kernel width and data increase on stability and fidelity criteria have been performed. Non-dependence on kernel width in fidelity is also one of the strengths of the proposed method.

Improvement of Machine Learning-Based Modelling of Container Ship’s Main Particulars with Synthetic Data

One of the main problems in the application of machine learning techniques is the need for large amounts of data necessary to obtain a well-generalizing model. This is exacerbated for studies in which it is not possible to access large amounts of data—for example, in the case of ship main data modelling, where a limited amount of real-world data (ship main data) is available for dataset creation. In this paper, a synthetic data generation technique has been applied to generate a large amount of synthetic data points regarding container ships’ main particulars. Models are trained using a multilayer perceptron (MLP) regressor on both original and synthetic data mixed with original data points. Then, the authors validate the performance of the obtained models on the original data and conclude whether a synthetic-data-based approach can be used to develop models in instances where the amount of data on ship main particulars may be limited. The results demonstrate an improvement across almost all outputs, ranging between 0.01 and 0.21 when evaluated using the coefficient of determination (R2) and between 0.27% and 3.43% when models are evaluated with mean absolute percentage error (MAPE). This indicates that the application of synthetic data can indeed be used for the improvement of ML-based model performance. The presented study demonstrates that the application of ML-based syncretization techniques can provide significant improvements to the process of ML-based determination of a ship’s main particulars at the early design stage. This paper suggests that, in cases where only a small dataset is available, artificial neural networks (ANN) can still be effectively employed to derive early-stage design values for the main particulars through the use of synthetic data.

Optimization method of the cam curve for acontinuous zoom optical system.

To make the driving force of a zoom cam cylinder stable and easy to control, the correlation between the cam curves of the zoom group and the compensation group is established effectively by reasonably selecting and arranging the coordinate system.Two optimization methods of cam curve are proposed. First, the original data are processed directly by the function of the target cam curve, and then the dynamic parameter such as the pressure angle is verified again to obtain the expected cam curve. This optimization method is simple and convenient, and a variety of optimization results can be constructed. Second, a physical model of multiple cam curves and the driving force with only one variable is constructed. The step search optimization method is used to process the original data point by point, and the best cam curve matching the model is obtained. Through dynamic simulation, it is determined that the cam curves designed by this optimization method can drive the zoom group and compensation group to complete the whole continuous zoom movement with a stable torque.

Efficient Computation of Spatial Entropy Measures.

Entropy indices are commonly used to evaluate the heterogeneity of spatially arranged data by exploiting various approaches capable of including spatial information. Unfortunately, in practical studies, difficulties can arise regarding both the availability of computational tools for fast and easy implementation of these indices and guidelines supporting the correct interpretation of the results. The present work addresses such issues for the most known spatial entropy measures: the approach based on area partitions, the one based on distances between observations, and the decomposable spatial entropy. The newly released version of the R package SpatEntropy is introduced here and we show how it properly supports researchers in real case studies. This work also answers practical questions about the spatial distribution of nesting sites of an endangered species of gorillas in Cameroon. Such data present computational challenges, as they are marked points in continuous space over an irregularly shaped region, and covariates are available. Several aspects of the spatial heterogeneity of the nesting sites are addressed, using both the original point data and a discretised pixel dataset. We show how the diversity of the nesting habits is related to the environmental covariates, while seemingly not affected by the interpoint distances. The issue of scale dependence of the spatial measures is also discussed over these data. A motivating example shows the power of the SpatEntropy package, which allows for the derivation of results in seconds or minutes with minimum effort by users with basic programming abilities, confirming that spatial entropy indices are proper measures of diversity.

Application of k-sigma Algorithm Based on Monte Carlo in Outlier Detection

Under the background of intelligent times based on information technology, the analysis and replacement of outliers become particularly critical for databases in the face of massive sequential data streams. In order to improve the effectiveness and practicability of the detection method, this paper determines the abnormal scores of data points in a specific data set by using the K-Sigma algorithm of the Monte Carlo method, adaptively adjusts the k value according to the abnormal scores, and marks the abnormal points.The advantages of the traditional k-sigma algorithm are fast operation speed and theoretical basis, while the disadvantages are that each index dimension is determined independently. However, in industrial production, due to the multi-index dimension of time series data, multiple size indicators of a data set are related to each other. Therefore, it is necessary to comprehensively consider whether an anomaly is abnormal by combining it with the data of other indicators. In addition, when there are no outliers in the data set, the traditional k-sigma algorithm is used to take the mean and variance of the data set as parameters. Due to the limitations of the data set itself, some data points will be mislabeled as outliers. Through the k-sigma algorithm based on the Monte Carlo method, we can effectively solve the above problems.The k-sigma algorithm based on the Monte Carlo method can generate a normal distribution according to the distribution of original data points, and extract a large quantity of data from the distribution to generate Monte Carlo data set. The Mahalanobis distance li from each sample point to the mean in the Monte Carlo data set is calculated and compared with the Mahalanobis distance lj from the samples to be detected to the mean in the original data set. According to the number of sample points satisfying li < lj in the Monte Carlo data set, the value of the parameter kj is determined adaptively, and thus the outliers are determined. We implemented the k-sigma algorithm based on the Monte Carlo method through python, evaluated the effectiveness of the algorithm by accuracy, recall rate, and F1 score, and compared it with some machine learning algorithms. The results verified the feasibility and effectiveness of the algorithm, which can be used for real-time anomaly detection in the energy management database.

Priori Anchor Labels Supervised Scalable Multi-View Bipartite Graph Clustering

Although multi-view clustering (MVC) has achieved remarkable performance by integrating the complementary information of views, it is inefficient when facing scalable data. Proverbially, anchor strategy can mitigate such a challenge a certain extent. However, the unsupervised dynamic strategy usually cannot obtain the optimal anchors for MVC. The main reasons are that it does not consider the fairness of different views and lacks the priori supervised guidance. To completely solve these problems, we first propose the priori anchor graph regularization (PAGG) for scalable multi-view bipartite graph clustering, dubbed as SMGC method. Specifically, SMGC learns a few representative consensus anchors to simulate the numerous view data well, and constructs a bipartite graph to bridge the affinities between the anchors and original data points. In order to largely improve the quality of anchors, PAGG predefines prior anchor labels to constrain the anchors with discriminative cluster structure and fair view allocation, such that a better bipartite graph can be obtained for fast clustering. Experimentally, abundant of experiments are accomplished on six scalable benchmark datasets, and the experimental results fully demonstrate the effectiveness and efficiency of our SMGC.

PrimeNet: Pre-training for Irregular Multivariate Time Series

Real-world applications often involve irregular time series, for which the time intervals between successive observations are non-uniform. Irregularity across multiple features in a multi-variate time series further results in a different subset of features at any given time (i.e., asynchronicity). Existing pre-training schemes for time-series, however, often assume regularity of time series and make no special treatment of irregularity. We argue that such irregularity offers insight about domain property of the data—for example, frequency of hospital visits may signal patient health condition—that can guide representation learning. In this work, we propose PrimeNet to learn a self-supervised representation for irregular multivariate time-series. Specifically, we design a time sensitive contrastive learning and data reconstruction task to pre-train a model. Irregular time-series exhibits considerable variations in sampling density over time. Hence, our triplet generation strategy follows the density of the original data points, preserving its native irregularity. Moreover, the sampling density variation over time makes data reconstruction difficult for different regions. Therefore, we design a data masking technique that always masks a constant time duration to accommodate reconstruction for regions of different sampling density. We learn with these tasks using unlabeled data to build a pre-trained model and fine-tune on a downstream task with limited labeled data, in contrast with existing fully supervised approach for irregular time-series, requiring large amounts of labeled data. Experiment results show that PrimeNet significantly outperforms state-of-the-art methods on naturally irregular and asynchronous data from Healthcare and IoT applications for several downstream tasks, including classification, interpolation, and regression.

Parameter Identification of Bivariate Fractal Interpolation Surfaces by Using Convex Hulls

The scope of this article is to identify the parameters of bivariate fractal interpolation surfaces by using convex hulls as bounding volumes of appropriately chosen data points so that the resulting fractal (graph of) function provides a closer fit, with respect to some metric, to the original data points. In this way, when the parameters are appropriately chosen, one can approximate the shape of every rough surface. To achieve this, we first find the convex hull of each subset of data points in every subdomain of the original lattice, calculate the volume of each convex polyhedron and find the pairwise intersections between two convex polyhedra, i.e., the convex hull of the subdomain and the transformed one within this subdomain. Then, based on the proposed methodology for parameter identification, we minimise the symmetric difference between bounding volumes of an appropriately selected set of points. A methodology for constructing continuous fractal interpolation surfaces by using iterated function systems is also presented.

Nonlocal, local and global preserving stacked autoencoder based fault detection method for nonlinear process monitoring

In recent years, the application of deep learning methods has improved the ability of nonlinear feature extraction in industrial processes. However, most of the deep learning methods cannot consider the manifold structure-related information (i.e., nonlocal, local and global manifold structure) of the data. To extract the vital structure-related features, a novel local, nonlocal and global preserving stacked autoencoder (NLGPSAE) for nonlinear process monitoring is proposed. NLGPSAE has an ability to extract crucial structure-related features by introducing a new regularized objective function with local, nonlocal and global structural information. For the preservation of local structure, NLGPSAE can preserve the original neighbor data points to be neighbor data points in the reconstructed space; for the preservation of nonlocal structures, NLGPSAE projects the nonlocal data points to be far apart in the reconstructed space; for the preservation of the global structure, the distance relationship between the original data points and the center of the data points is preserved in the reconstructed space. Two statistics Hotelling's T-squared (T2) and squared prediction error (SPE) based on features extracted by NLGPSAE are established for fault detection. The effectiveness of the proposed algorithm is verified in a complex numerical process and Tennessee Eastman process.

Non-uniform subdivision schemes of ωB-spline curves and surfaces with variable parameters

In this paper, we propose a new non-uniform subdivision scheme that includes a tension parameters sequence. Each tension parameter of the sequence is assigned at each edge of the initial control polygon. The proposed scheme can produce limiting curves that are more consistent with the original data points and the control polygon. It has also the advantage of generating a wide variety of shapes for the limiting curves. Moreover, to produce smooth limit surfaces, the proposed scheme is firstly extended to tensor-product surfaces, then to meshes of arbitrary topology. The convergence and smoothness of the proposed scheme are proven. Numerical results that illustrate the advantages of the proposed non-uniform scheme are given. • This family of non-uniform subdivision schemes accompanied with algorithms for curves and surfaces can be used in many engineering applications. • It is able to generate a wide variety of shapes for the limiting curves and surfaces as it includes a tension parameters sequence. • It allows local control of the shape of the resulting limit curves and surfaces. • It is extended to meshes of arbitrary topology to generate complex surface subdivisions. • It turns out to be the uniform scheme, when the considered tension parameters in the sequence are all equals. • It turns out to be several well-known subdivision schemes, such as Doo–Sabin and Catmull–Clark subdivision, when all the tension parameters in the sequence are equals to 1.

An Implementation of Galactic White Dwarf Binary Data Analysis for MLDC-3.1

The space-borne gravitational wave detectors will observe a large population of double white dwarf binaries in the Milky Way. However, the search for double white dwarfs in the gravitational wave data will be time-consuming due to the large number of templates involved and antenna response calculation. In this paper, we implement an iterative combinatorial algorithm to search for double white dwarfs in MLDC-3.1 data. To quickly determine the rough parameters of the target sources, the following algorithms are adopted in a coarse search process: (1) using the downsampling method to reduce the number of original data points; (2) using the undersampling method to speed up the generation of a single waveform template; (3) using the stochastic template bank method to quickly construct the waveform template bank while achieving high coverage of the parameter space; (4) combining the FFT acceleration algorithm with the stochastic template bank to reduce the calculation time of a single template. A fine search process is applied to further determine the parameters of the signals based on the coarse search, for which we adopt the particle swarm optimization. Finally, we detect double white dwarf signals, validating the feasibility of our method.

Center consistency guided multi-view embedding anchor learning for large-scale graph clustering

Multi-view clustering has attracted extensive attention since it can integrate the complementary information of different views. Nonetheless, most existing methods suffer from cubic time computational complexity, so they have to face the challenge of maladjustment to medium and large-scale datasets. Whereupon, researchers usually employ anchor selection strategies to select a proportion of data points as landmarks to simulate the original data, and then build a bipartite graph to bridge the relationship between the original data points and anchors. However, for multi-view data, the existing strategies have three shortcomings, including: (1) on account of redundancy and corruption, the quality of anchors may be greatly affected; (2) these strategies are static ones or deficiency of views consistency; and (3) due to the differences in source, latent distribution, and the dimension of each view, the consensus bipartite graph cannot exploit the comprehensive information of all views precisely. In order to address these problems, this paper first proposes a novel method for scalable clustering, dubbed center consistency guided multi-view embedding anchor learning (2C-MEAL). Specifically, considering the redundancy and corruption, embedding learning is adopted to filter out the adverse information, as well as unify the dimensions of all views. Then, dynamic center consistency guided anchor learning is proposed to adaptively learn a set of high-quality anchors in the clean embedding space. Meanwhile, the refined bipartite graph is obtained for clustering purposes. Overall, 2C-MEAL has promising clustering performance and linear complexity. Experimentally, clustering experiments on benchmark datasets demonstrate that 2C-MEAL possesses effectiveness and efficiency over state-of-the-art methods.

Deep ensemble with Neural Networks to model power curve uncertainty

Renewable energy is an essential driver towards tackling the climate crisis. Given the increasing need to ensure sustainability and reduce greenhouse gas emissions, wind energy has become one of the most relevant areas to consider in today’s society. However, it appears challenging to rely on this type of energy because of the uncertain nature of the wind. It is essential to use monitoring tools that can accurately capture the uncertainty of the underlying environmental processes and estimate the generated power to ensure the reliability of wind-based power systems. One of these monitoring tools is the power curve, which captures the relationship between wind speed and expected power production. Many regression techniques have previously been used to predict this power production, including Artificial Neural Networks (ANN). The problem is that they do not capture the uncertainty of the predicted power production. In contrast with the power curve models already implemented when using ANNs, we propose a novel approach based on deep ensembles to predict the mean and model variance on the expected power production. We used three months of Supervisory Control and Data Acquisition data from more than 40 turbines to set up the experiments. The results show that the mean and standard deviation fit the data well. It also appears that the expected mean curve goes through the center of the cloud of original data points, while the standard deviation accurately captures the spread of the data. We conclude that that the deep ensemble approach provides an accurate prediction mechanism to estimate the power production from the wind speed while also offering an uncertainty measure. Therefore, the deep ensemble approach allows us to construct a power curve with uncertainty information.

An Exploration of Consistency Learning with Data Augmentation

&#x0D; &#x0D; &#x0D; Deep Learning has achieved remarkable success with Supervised Learning. Nearly all of these successes require very large manually annotated datasets. Data augmentation has enabled Supervised Learning with less labeled data, while avoiding the pitfalls of overfitting. However, Supervised Learning still fails to be Robust, making different predictions for original and augmented data points. We study the addition of a Consistency Loss between representations of original and augmented data points. Although this offers additional structure for invariance to augmentation, it may fall into the trap of representation collapse. Representation collapse describes the solution of mapping every input to a constant output, thus cheating to solve the consistency task. Many techniques have been developed to avoid representation collapse such as stopping gradients, entropy penalties, and applying the Consis- tency Loss at intermediate layers. We provide an analysis of these techniques in interaction with Supervised Learning for the CIFAR-10 image classification dataset. Our consistency learning models achieve a 1.7% absolute improvement on the CIFAR-10 original test set over the supervised baseline. More interestingly, we are able to dramatically reduce our proposed Distributional Distance metric with the Consistency Loss. Distributional Distance provides a more fine- grained analysis of the invariance to corrupted images. Readers will understand the practice of adding a Consistency Loss to improve Robustness in Deep Learning.&#x0D; &#x0D; &#x0D;

TLMOTE: A Topic-based Language Modelling Approach for Text Oversampling

Training machine learning and deep learning models on unbalanced datasets can lead to a bias portrayed by the models towards the majority classes. To tackle the problem of bias towards majority classes, researchers have presented various techniques to oversample the minority class data points. Most of the available state-of-the-art oversampling techniques generate artificial data points which cannot be comprehensibly understood by the reader, despite the synthetic data points generated being similar to the original minority class data points. In this work, we present Topic-based Language Modelling Approach for Text Oversampling (TLMOTE), a novel text oversampling technique for supervised learning from unbalanced datasets. TLMOTE improves upon previous approaches by generating data points which can be intelligibly understood by the reader, can relate to the main topics of the minority class, and introduces more variations to the synthetic data generated. We evaluate the efficacy of our approach on various tasks like Suggestion Mining SemEval 2019 Task 9 Subtasks A and B, SMS Spam Detection, and Sentiment Analysis. Experimental results verify that oversampling unbalanced datasets using TLMOTE yields a higher macro F1 score than with other oversampling techniques.

Sand-mudstone modeling of fluvial fan sedimentary facies: a case study of Shanxi Formation reservoir in Ordos Basin

The sand bodies formed by braided fluvial fan deposits have a certain distinctiveness. They not only have the characteristics of fluvial facies sandbodies but also follow the distribution law of alluvial fan sand bodies. The variation law of sandbodies that are present along and perpendicular to a channel is relatively complex. Therefore, constraints in the modeling process of sand–mudstone facies of this type of reservoir are essential. This study selects the second member of the Shanxi Formation reservoir formed by a braided fluvial fan in the middle of Ordos Basin to perform sand–mudstone facies modeling. First, by studying the lithology and sedimentary structure of the area, the sedimentary characteristics and sand body distribution law of braided river fan are analyzed. Then, the original data points are analyzed, the variation function with high convergence is obtained, and the sand–mud facies model under the constraint of sedimentary facies is established using the random modeling method. Finally, the accuracy of the established random model is tested via single-well thinning, multi-well thinning, and random seed model similarity. The test results confirm that the distribution law of the sand and mudstone in the model is highly similar to that of the actual stratum. And it also conforms to the sedimentary model of braided fluvial fan. The accuracy of the model established by this method is reliable, and the method can be used to predict sand body distribution in areas with low well pattern density.

Computation of quantile sets for bivariate ordered data

Algorithms are proposed for the computation of set-valued quantiles and the values of the lower cone distribution function for bivariate data sets. These new objects make data analysis possible involving an order relation for the data points in form of a vector order in two dimensions. The bivariate case deserves special attention since two-dimensional vector orders are much simpler to handle than such orders in higher dimensions. Several examples illustrate how the algorithms work and what kind of conclusions can be drawn with the proposed approach. As a new feature, it is observed that the computational effort depends on how much the original data points are aligned with respect to the vector order.

Density Peaks Clustering Based on Feature Reduction and Quasi-Monte Carlo

Density peaks clustering (DPC) is a well-known density-based clustering algorithm that can deal with nonspherical clusters well. However, DPC has high computational complexity and space complexity in calculating local density ρ and distance δ , which makes it suitable only for small-scale data sets. In addition, for clustering high-dimensional data, the performance of DPC still needs to be improved. High-dimensional data not only make the data distribution more complex but also lead to more computational overheads. To address the above issues, we propose an improved density peaks clustering algorithm, which combines feature reduction and data sampling strategy. Specifically, features of the high-dimensional data are automatically extracted by principal component analysis (PCA), auto-encoder (AE), and t-distributed stochastic neighbor embedding (t-SNE). Next, in order to reduce the computational overhead, we propose a novel data sampling method for the low-dimensional feature data. Firstly, the data distribution in the low-dimensional feature space is estimated by the Quasi-Monte Carlo (QMC) sequence with low-discrepancy characteristics. Then, the representative QMC points are selected according to their cell densities. Next, the selected QMC points are used to calculate ρ and δ instead of the original data points. In general, the number of the selected QMC points is much smaller than that of the initial data set. Finally, a two-stage classification strategy based on the QMC points clustering results is proposed to classify the original data set. Compared with current works, our proposed algorithm can reduce the computational complexity from O n 2 to O N n , where N denotes the number of selected QMC points and n is the size of original data set, typically N ≪ n . Experimental results demonstrate that the proposed algorithm can effectively reduce the computational overhead and improve the model performance.

Graph Embedding With Data Uncertainty

Spectral-based subspace learning is a common data preprocessing step in many machine learning pipelines. The main aim is to learn a meaningful low dimensional embedding of the data. However, most subspace learning methods do not take into consideration possible measurement inaccuracies or artifacts that can lead to data with high uncertainty. Thus, learning directly from raw data can be misleading and can negatively impact the accuracy. In this paper, we propose to model artifacts in training data using probability distributions; each data point is represented by a Gaussian distribution centered at the original data point and having a variance modeling its uncertainty. We reformulate the Graph Embedding framework to make it suitable for learning from distributions and we study as special cases the Linear Discriminant Analysis and the Marginal Fisher Analysis techniques. Furthermore, we propose two schemes for modeling data uncertainty based on pair-wise distances in an unsupervised and a supervised contexts.

Estimating Weak Lensing Convergence Correlation of Type-Ia Supernovae From 5-Year Snls Data by Internal Error Estimate Technique

We report non-zero weak lensing convergence correlation signal of Type-Ia supernovae from 5-year Supernovae Legacy Survey data. For our analysis we utilize 296 supernovae magnification data from 5-year SNLS in the weak lensing limit. The data we use consists of measurements from four different patches, each covering 1 square degree of the sky, merged together. We demonstrate that it is possible to have a very good estimate of the two point correlation function from this data using internal error estimate technique. In order to have a good estimate of the corresponding covariance matrix we apply bootstrap spatial re-sampling technique where we reshuffle the original data consisting of 296 data points 100-10000 times and compare the results with that obtained from original data points. We show that this technique helps us arrive at a reliable conclusion on weak lensing convergence even though the original dataset comprises of a small number of data points. This also allows us to compute the corresponding covariance matrix with great accuracy.