Class Boundaries Research Articles

Multiclass Fuzzily Weighted Adaptive-Boosting-Based Self-Organizing Fuzzy Inference Ensemble Systems for Classification

Adaptive boosting (AdaBoost) is a widely used technique to construct a stronger ensemble classifier by combining a set of weaker ones. Zero-order fuzzy inference systems (FISs) are very powerful prototype-based predictive models for classification, offering both great prediction precision and high user interpretability. However, the use of zero-order FISs as base classifiers in AdaBoost has not been explored yet. To bridge the gap, in this article, a novel multiclass fuzzily weighted AdaBoost (FWAdaBoost)-based ensemble system with a self-organizing fuzzy inference system (SOFIS) as the ensemble component is proposed. To better incorporate the SOFIS, FWAdaBoost utilizes the confidence scores produced by the SOFIS in both sample weight updating and ensemble output generation, resulting in more accurate classification boundaries and greater prediction precision. Numerical examples on a wide range of benchmark classification problems demonstrate the efficacy of the proposed approach.

Generation of Controlled Synthetic Samples and Impact of Hyper-Tuning Parameters to Effectively Classify the Complex Structure of Overlapping Region

The classification of imbalanced and overlapping data has provided customary insight over the last decade, as most real-world applications comprise multiple classes with an imbalanced distribution of samples. Samples from different classes overlap near class boundaries, creating a complex structure for the underlying classifier. Due to the imbalanced distribution of samples, the underlying classifier favors samples from the majority class and ignores samples representing the least minority class. The imbalanced nature of the data—resulting in overlapping regions—greatly affects the learning of various machine learning classifiers, as most machine learning classifiers are designed to handle balanced datasets and perform poorly when applied to imbalanced data. To improve learning on multi-class problems, more expertise is required in both traditional classifiers and problem domain datasets. Some experimentation and knowledge of hyper-tuning the parameters and parameters of the classifier under consideration are required. Several techniques for learning from multi-class problems have been reported in the literature, such as sampling techniques, algorithm adaptation methods, transformation methods, hybrid methods, and ensemble techniques. In the current research work, we first analyzed the learning behavior of state-of-the-art ensemble and non-ensemble classifiers on imbalanced and overlapping multi-class data. After analysis, we used grid search techniques to optimize key parameters (by hyper-tuning) of ensemble and non-ensemble classifiers to determine the optimal set of parameters to enhance the learning from a multi-class imbalanced classification problem, performed on 15 public datasets. After hyper-tuning, 20% of the dataset samples are synthetically generated to add to the majority class of each respective dataset to make it more overlapped (complex structure). After the synthetic sample’s addition, the hyper-tuned ensemble and non-ensemble classifiers are tested over that complex structure. This paper also includes a brief description of tuned parameters and their effects on imbalanced data, followed by a detailed comparison of ensemble and non-ensemble classifiers with the default and tuned parameters for both original and synthetically overlapped datasets. We believe that the underlying paper is the first kind of effort in this domain, which will furnish various research aspects to with a greater focus on the parameters of the classifier in the field of learning from imbalanced data problems using machine-learning algorithms.

A multimetric macroinvertebrate index for the assessment of the ecological status of Lithuanian rivers

We present the design of the Lithuanian River Macroinvertebrate Index (LRMI) – a Water Framework Directive compliant method for assessing the ecological status of Lithuanian rivers that primarily addresses the eutrophication pressure. It is a multimetric index that averages the ecological quality ratios of two conventional metrics – the Danish Stream Fauna Index (DSFI) and the Average Score Per Taxon (ASPT) – as well as the two newly assembled metrics – the taxonomic richness of Diptera, Ephemeroptera, and Plecoptera (#DEP) and the difference between the pooled relative abundances of Ephemeroptera, Heteroptera, and Plecoptera versus Crustacea and Hirudinea (%EHP–%CrHi). The LRMI showed significant correlations with hydrochemical indicators of nutrient enrichment in the national river dataset. Moreover, our analysis showed its potential sensitivity toward hydromorphological alterations of the riverine habitats. Finally, the results indicated that further developments concerning minor type-specific adjustments of the currently set ecological class boundaries may be beneficial to the new method.

Machine learning the real discriminant locus

Parameterized systems of polynomial equations arise in many applications in science and engineering with the real solutions describing, for example, equilibria of a dynamical system, linkages satisfying design constraints, and scene reconstruction in computer vision. Since different parameter values can have a different number of real solutions, the parameter space is decomposed into regions whose boundary forms the real discriminant locus. This article views locating the real discriminant locus as a supervised classification problem in machine learning where the goal is to determine classification boundaries over the parameter space, with the classes being the number of real solutions. This article presents a novel sampling method which carefully samples a multidimensional parameter space. At each sample point, homotopy continuation is used to obtain the number of real solutions to the corresponding polynomial system. Machine learning techniques including nearest neighbors, support vector classifiers, and neural networks are used to efficiently approximate the real discriminant locus. One application of having learned the real discriminant locus is to develop a real homotopy method that only tracks real solution paths unlike traditional methods which track all complex solution paths. Examples show that the proposed approach can efficiently approximate complicated solution boundaries such as those arising from the equilibria of the N=4 Kuramoto model which was previously intractable using traditional methods.

Dialectical relation of temporal domination and class exploitation in Marx’s value theory

In this article, I demonstrate that temporal domination over subjects and producers in capitalist society is Marx’s central category in the critique of political economy. Hartmut Rosa who is well known in the critical tradition for his works on social acceleration refers to Marx’s value theory to explain the economic motor of social acceleration in the sphere of production and distribution, but he does not reach the very depth of Marx’s critique of abstract time domination and considered it as a subordinate and marginal aspect of Marx’s theory. I argue that Marx’s critique of class exploitation and commodification of social relations, in the final analysis, consists in the critique of domination of time over man. In particular, I articulate their relationship by representing the Hegelian onto-logical couple of essence and appearance; it is demonstrated that class exploitation as an internal and substantial structure of capitalism posits the temporal domination as an form of appearance which presupposes the first. Moreover, this article shows that class domination has taken a more inverted and mysterious form in digital capitalism through new time appropriation strategies. This implies that in contrast to Rosa who considers social acceleration above and beyond class boundaries, it still reinforces class exploitation and acts through it.

Depression detection based on linear and nonlinear speech features in I-vector/SVDA framework

This study proposes depression detection systems based on the i-vector framework for classifying speakers as depressed or healthy and predicting depression levels according to the Beck Depression Inventory-II (BDI-II). Linear and non-linear speech features are investigated as front-end features to i-vectors. To take advantage of the complementary effects of features, i-vector systems based on linear and non-linear features are combined through the decision-level fusion. Variability compensation techniques, such as Linear Discriminant Analysis (LDA) and Within-Class Covariance Normalization (WCCN), are widely used to reduce unwanted variabilities. A more generalizable technique than the LDA is required when limited training data are available. We employ a support vector discriminant analysis (SVDA) technique that uses the boundary of classes to find discriminatory directions to address this problem. Experiments conducted on the 2014 Audio-Visual Emotion Challenge and Workshop (AVEC 2014) depression database indicate that the best accuracy improvement obtained using SVDA is about 15.15% compared to the uncompensated i-vectors. In all cases, experimental results confirm that the decision-level fusion of i-vector systems based on three feature sets, TEO-CB-Auto-Env+Δ, Glottal+Δ, and MFCC+Δ+ΔΔ, achieves the best results. This fusion significantly improves classifying results, yielding an accuracy of 90%. The combination of SVDA-transformed BDI-II score prediction systems based on these three feature sets achieved RMSE and MAE of 8.899 and 6.991, respectively, which means 29.18% and 30.34% improvements in RMSE and MAE, respectively, over the baseline system on the test partition. Furthermore, this proposed combination outperforms other audio-based studies available in the literature using the AVEC 2014 database.

The Impact of Social Media on Social Cohesion in South Africa

Social cohesion is an important policy issue with social, political and economic consequences in South Africa. Given the country’s diverse population comprised of different races, ethnicities and migrants from around the world, its history of violence and the persistent incidences of xenophobia, social cohesion has become a prominent issue. The study foregoes discussing the challenges of achieving social cohesion to examine the nexus between social media and social cohesion promotion. Since lockdown was introduced in March 2020, social media has become a means for daily communication, bridging racial, ethnic, class and geographical boundaries. Against this background, this study explored the impact of social media on social cohesion in South Africa? The main objective was to examine the role played by social media in fostering social cohesion. A systematic review of relevant academic literature and research reports on social media and social cohesion was adopted in gathering data. By examining the social media social cohesion nexus, an illustration of social media as an essential determinant of social cohesion can be delineated. Teasing out these aspects is vital in enriching the current literature on understanding social cohesion. The study found that social media has become an essential platform for nation-building, potentially bridging racial, ethnic, cultural, and geographical gaps in South African society. However, due to the threats of misinformation and insensitive use, social media can hamstring social cohesion and cause instability. The study recommends that social media be used appropriately and sensitively in strengthening interpersonal relationships that transcend race, culture, ethnicity, nationality, and religion.

A novel method of quantitative evaluation and comprehensive classification of low permeability-tight oil reservoirs: A case study of Jidong Oilfield, China

The classification of low permeability-tight reservoirs is the premise of development. The deep reservoir of Shahejie 3 member contains rich low permeability-tight reserves, but the strong heterogeneity and complex micro pore structure make the main controlling factors subjective and the classification boundaries unclear. Therefore, a new indicator considering the interaction between fluid and rock named Threshold Flow Zone Indicator (TFZI) is proposed, it can be used as the main sequence of correlation analysis to screen the main controlling factors, and the clustering algorithm is optimized combined with probability distribution to determine the classification boundaries. The sorting coefficient, main throat radius, movable fluid saturation and displacement pressure are screened as the representative parameters for the following four key aspects: rock composition, microstructure, flow capacity and the interaction between rock and fluid. Compared with the traditional probability distribution and clustering algorithm, the boundary of the optimized clustering algorithm proposed in this paper is more accurate. The classification results are consistent with sedimentary facies, oil levels and oil production intensity. This method provides an important basis for the development of low permeability-tight reservoirs.

Exploring clusters of rare events using unsupervised random forests

Given highly imbalanced data, most learning algorithms face the challenge of accurately predicting rare events, while such cases are the ones that carry importance and useful knowledge. In a binary class label dataset, the rare events are the ones in the minority class. This study used a stroke dataset with a binary class label and the class imbalance ratio was 54:1. In addition to that, the dataset contained missing values and mixed data types. To identify the intrinsic structures in the minority class (the stroke group), Random Forest Clustering was used to produce the proximity matrix and fed to Partition around Medoid (PAM) clustering method to identify the optimal number of clusters. The proximity plot seems to show there could be cluster tendency as Hopkins’s statistics test value was H = 0.8735 and k=2 was identified to be the optimal number of clusters. Based on the internal cluster validation, however, the silhouette coefficient width was small (0.1), indicating that many of the data objects were within the other boundary of the other class. We have suggested a further investigation plan in this paper for the next action.

Distribution regularized self-supervised learning for domain adaptation of semantic segmentation

This paper proposes a novel pixel-level distribution regularization scheme (DRSL) for self-supervised domain adaptation of semantic segmentation. In a typical setting, the classification loss forces the semantic segmentation model to greedily learn the representations that capture inter-class variations in order to determine the decision (class) boundary. Due to the domain shift, this decision boundary is unaligned in the target domain, resulting in noisy pseudo labels adversely affecting self-supervised domain adaptation. To overcome this limitation, along with capturing inter-class variation, we capture pixel-level intra-class variations through class-aware multi-modal distribution learning (MMDL). Thus, the information necessary for capturing the intra-class variations is explicitly disentangled from the information necessary for inter-class discrimination. Features captured thus are much more informative, resulting in pseudo-labels with low noise. This disentanglement allows us to perform separate alignments in discriminative space and multi-modal distribution space, using cross-entropy based self-learning for the former. For later, we propose a novel stochastic mode alignment method, by explicitly decreasing the distance between the target and source pixels that map to the same mode. The distance metric learning loss, computed over pseudo-labels and backpropagated from multi-modal modeling head, acts as the regularizer over the base network shared with the segmentation head. The results from comprehensive experiments on synthetic to real domain adaptation setups, i.e., GTA-V/SYNTHIA to Cityscapes, show that DRSL outperforms many existing approaches (a minimum margin of 2.3% and 2.5% in mIoU for SYNTHIA to Cityscapes).

A novel SVM-based decision framework considering feature distribution for Power Transformer Fault Diagnosis

International Electrotechnical Commission (IEC) proposed the IEC three-ratio method based on Dissolved Gas Analysis (DGA), which is one of the most effective tools for Power Transformer Fault Diagnosis (PTFD). However, the PTFD accuracy is generally limited because the classification boundary could be too stiff to classify samples located near the boundary. The Support Vector Machine (SVM) was applied to PTFD to improve diagnosis accuracy, while traditional SVM multi-classification methods and parameter optimization algorithms are subject to poor training efficiency. As a result, the SVM-based PTFD model is difficult to update frequently with the accumulation of fault data. A new SVM-based PTFD decision framework is proposed in this paper which can significantly boost the training efficiency and ensure the accuracy. In the proposed framework, a multi-step feature extraction process consisting of characteristic gas concentration and its ratios is applied. Based on the feature distribution of various samples, a proper SVM multi-classification method is presented using a hierarchical decision tree structure. In addition, according to the principles of SVM and radial basis kernel function, a Support Vector feature-based parameter optimization algorithm (SVFB) is proposed. IEC TC 10 data and the historical data of online transformer monitoring provided by the State Grid Corporation of China are adopted as sample sets. The simulation results demonstrate that the proposed decision framework can reach high diagnosis accuracy while shortening the training time.

Gaussian gravitation for cluster ensembles

Gravity-based clustering methods can effectively distinguish the differences between the data points close to the center of a dataset and those on a class boundary. However, most existing methods are only suitable for processing real-valued data. In this paper, the processing scope of gravitational models is extended to discrete data and then utilized to solve cluster ensemble problems. First, a novel Gaussian gravitational model for cluster ensemble (GGMCE) is proposed in which each base cluster can be transformed into an object with a mass via Newton’s second law of motion and thus extending gravitational models to discrete data. Furthermore, instead of setting hyperparameters for the number of neighbors, Gaussian agency strategy is proposed to explore the final cluster assignment. To validate the capability of the proposed model, extensive experiments on a variety of real-world datasets are conducted and the result demonstrates the superiority of the proposed model over the state-of-the-art models.

Context of racialized assimilation: a case of second-generation Somali-Canadians

ABSTRACT Can second-generation racialized Canadians cross racial and class boundaries into middle-class mainstream society? Currently, neo-assimilation theory anticipates identificational and socioeconomic assimilation into the mainstream, while segmented-assimilation theory argues that racial and economic structures inhibit racialized groups’ assimilation. An emerging strand of assimilation theory – racialized incorporation – hypothesizes that, for racialized individuals, higher sociocultural capital leads to socioeconomic upward mobility, while their identificational and socioeconomic assimilation remain constrained by racial hierarchies. We draw on 118 qualitative interviews with second-generation Somali-Canadians to assess which of these perspectives best describes their subjective experiences of assimilation in Canada. Though we find support for the racialized assimilation hypothesis, our participants’ socioeconomic achievements exceed the predictions of this perspective. We, therefore, argue that, under Canadian multiculturalism policies and ideology, race is an attitudinal hurdle to navigate rather than a structural barrier against assimilation. We invite future research to consider the contextualized effects of race on assimilation.

Zero-Bias Deep-Learning-Enabled Quickest Abnormal Event Detection in IoT

Abnormal event detection with the lowest latency is an indispensable function for safety-critical systems, such as cyber defense systems. However, as systems become increasingly complicated, conventional sequential event detection methods become less effective, especially when we need to define indicator metrics from complicated data manually. Although deep neural networks (DNNs) have been used to handle heterogeneous data, the theoretic assurability and explainability are still insufficient. This article provides a holistic framework for the quickest and sequential detection of abnormalities and time-dependent abnormal events. We explore the latent space characteristics of zero-bias neural networks considering the classification boundaries and abnormalities. We then provide a novel method to convert zero-bias DNN classifiers into performance-assured binary abnormality detectors. Finally, we provide a sequential quickest detection (QD) scheme that provides the theoretically assured lowest abnormal event detection delay under false alarm constraints using the converted abnormality detector. We verify the effectiveness of the framework using real massive signal records in aviation communication systems and simulation. Codes and data are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/pcwhy/AbnormalityDetectionInZbDNN</uri> .

CMS2-Net: Semi-Supervised Sleep Staging for Diverse Obstructive Sleep Apnea Severity.

Although the development of computer-aided algorithms for sleep staging is integrated into automatic detection of sleep disorders, most supervised deep learning-based models might suffer from insufficient labeled data. While the adoption of semi-supervised learning (SSL) can mitigate the issue, the SSL models are still limited to the lack of discriminative feature extraction for diverse obstructive sleep apnea (OSA) severity. This model deterioration might be exacerbated during the domain adaptation. Such exploration on the alleviation of domain-shift of SSL model between different OSA conditions has attracted more and more attentions from the clinic. In this work, a co-attention meta sleep staging network (CMS2-net) is proposed to simultaneously deal with two issues: the inter-class disparity problem and the intra-class selection problem. Within CMS2-net, a co-attention module and a triple-classifier are designed to explicitly refine the coarse feature representations by identifying the class boundary inconsistency. Moreover, the mutual information with meta contrastive variance is introduced to supervise the gradient stream from a multi-scale view. The performance of the proposed framework is demonstrated on both public and local datasets. Furthermore, our approach achieves the state-of-the-art SSL results on both datasets.

Интерпретация результатов гранулометрического анализа почвы в разных школах почвоведения

The article discusses a few of today&amp;#x27;s applicable soil classifications in terms of their texture: the one by A. Kachinsky (Russians) and other two – by the US Department of Agriculture (USDA) and the World Reference Base for Soil Resources (WRB). The article also presents several discrepancies in the nature of the source of research (primary soil particles), the width of the borders of the key textural classes, and also in the need to take into account possible ways of their formation when classifying. Bearing it in mind, we have studied 57 separate soil samples of the Calcic Chernozem of the Rostov Region. The comparative study of the results of the particle-size distribution analysis was performed based on the results obtained with the help of the laser diffraction method, since it is not directly related to any of the above classifications, thus being free from whatever lasting stereotypes. The choice of options for the conditions of soil formation as presented in the Russian classification seems a key decisive parameter while interpreting experimental results. In the example under consideration, this classification allows for defining the variety of a textural class as clay (from light to heavy) depending on what type of soil formation the sample may be related to. Due to some discrepancies in determining the boundaries of the textural classes, we have observed redistribution of primary soil particles whereas their factual amount remains unchanged. All these differences are vital nowadays in view of globalization and boost in the number and importance of international research.

DeepVisualInsight: Time-Travelling Visualization for Spatio-Temporal Causality of Deep Classification Training

Understanding how the predictions of deep learning models are formed during the training process is crucial to improve model performance and fix model defects, especially when we need to investigate nontrivial training strategies such as active learning, and track the root cause of unexpected training results such as performance degeneration. In this work, we propose a time-travelling visual solution DeepVisualInsight (DVI), aiming to manifest the spatio-temporal causality while training a deep learning image classifier. The spatio-temporal causality demonstrates how the gradient-descent algorithm and various training data sampling techniques can influence and reshape the layout of learnt input representation and the classification boundaries in consecutive epochs. Such causality allows us to observe and analyze the whole learning process in the visible low dimensional space. Technically, we propose four spatial and temporal properties and design our visualization solution to satisfy them. These properties preserve the most important information when projecting and inverse-projecting input samples between the visible low-dimensional and the invisible high-dimensional space, for causal analyses. Our extensive experiments show that, comparing to baseline approaches, we achieve the best visualization performance regarding the spatial/temporal properties and visualization efficiency. Moreover, our case study shows that our visual solution can well reflect the characteristics of various training scenarios, showing good potential of DVI as a debugging tool for analyzing deep learning training processes.

Myanmar’s Coup d’état and the Struggle for Federal Democracy and Inclusive Government

This article reviews the first twelve months of the civil disobedience movement in Myanmar following the 1 February 2021 coup d’état and its many dynamics and manifestations. Myanmar’s ‘Spring Revolution’ generated a shared sense of national unity—overcoming gender, ethnic, religious and class boundaries, but raising questions about the long-term sustainability of nonviolent civil resistance in a state where the military has for decades wielded political and economic power. Since the coup, Myanmar has been in turmoil, paralysed by instability which escalated after the military’s deadly crackdown on pro-democracy activists. The article charts the growth of the Civil Disobedience Movement (CDM), its multiple methods of strategic resistance and non-cooperation, and the radicalisation of the resistance agenda. It analyses the formation of the Committee Representing the Pyidaungsu Hluttaw (CRPH), the creation of the interim National Unity Government (NUG), the founding of the National Unity Consultative Council (NUCC) and the inauguration of the People’s Defence Force (PDF). It examines the implications for Myanmar when the crisis reached a more complex phase after the military’s open use of force and terror on the broader civilian population prompted the NUG to declare war on the junta, and to urge ethnic armed organisations (EAOs) and newly formed anti-junta civilian militias (PDF) to attack the State Administration Council (SAC) as a terrorist organisation. The NUG now opposes the military junta by strategic and peaceful non-cooperation, armed resistance, and international diplomacy. This paper considers whether the predominantly nonviolent civil resistance movement’s struggle for federal democracy and inclusive governance is laying the foundations for eventual transition to a fully democratic future or whether the cycles of violence will continue as the military continues to control power by using intimidation and fear. It notes that the coup has destroyed the economy and expanded Myanmar’s human rights and humanitarian crises but has also provided the opportunity for Myanmar’s people to explore diverse visions of a free, federal, democratic and accountable Myanmar. It finally examines the possibilities for future peaceful nation building, reconciliation, and the healing of the trauma of civil war.

Ensemble mapping as an alternative to baseline seafloor sediment mapping and monitoring

Seafloor sediment mapping traditionally relies on the experience and expertise of practitioners to classify sediment classes based on acoustic backscatter data and ground-truth information. However, recent advancements in automated seafloor mapping present a major development in objective methods that offer practical application for seafloor mapping and monitoring campaigns. In this study, a class-specific approach of ensemble modelling (ensemble mapping) was used to classify the sediment classes of a large-scale seafloor area (1550 km2) in the western Sylt Outer Reef, German North Sea. A pixel-by-pixel comparison of the modelled map and manually digitized map was also conducted to assess the efficiency of the ensemble mapping approach. The resulting seafloor sediment map, with an overall accuracy of 73%, demonstrates five sediment classes that represent most of the seabed of the German North Sea. The manually classified and ensembled maps were 63% identical, but mismatches were observed in the transitional boundaries of soft sediment classes and in stony areas that were not predicted in manual classification. The inconsistencies between the two maps was attributed to the different interpretation of sediment boundaries, the simplification of the sediment classification scheme, and the ability of ensemble mapping to classify more areas than manual classification. This study found that ensemble mapping performs better in characterizing coarse materials and produces maps that are comparable to the maps produced by manual classification, while the production time and degree of subjectivity in the analysis are minimal. Hence, ensemble mapping is a viable alternative to create baseline seafloor sediment maps that can be used for environmental monitoring and resource planning.

Construction of Tourism Management Engineering Based on Data Mining Technology

In order to improve the construction quality of tourism management projects, this paper applies data mining algorithm to tourism management, and analyzes the SMOTE algorithm. According to the improvement direction, this paper proposes two improved algorithms, KM-SMOTE and RM-SMOTE, and uses the clustering algorithm to preprocess the minority data set. Moreover, on this basis, this paper establishes clusters and obtains cluster centers. The deficiencies of fuzzy positive and negative class boundaries can be effectively solved by oversampling with the cluster center as the base point, and in the case of appropriately expanding the reasonable data interpolation method, the area space of boundary interpolation can be shrunk, and the performance of the algorithm can be improved. It can be seen from the simulation test research that the tourism management system based on data mining proposed in this paper can play an important role in tourism management and effectively promote the improvement of tourism management efficiency.