Original Class Research Articles

Weighted support vector machine for extremely imbalanced data

Based on an asymptotically optimal weighted support vector machine (SVM) that introduces label shift, a systematic procedure is derived for applying oversampling and weighted SVM to extremely imbalanced datasets with a cluster-structured positive class. This method formalizes three intuitions: (i) oversampling should reflect the structure of the positive class; (ii) weights should account for both the imbalance and oversampling ratios; (iii) synthetic samples should carry less weight than the original samples. The proposed method generates synthetic samples from the estimated positive class distribution using a Gaussian mixture model. To prevent overfitting to excessive synthetic samples, different misclassification penalties are assigned to the original positive class, synthetic positive class, and negative class. The proposed method is numerically validated through simulations and an analysis of Republic of Korea Army artillery training data.

Degrees and Limits of Adverbial Transposition of Nouns in the Russian Language (Based on Formations with the Semantics of the Season and the Day)

The article presents a comprehensive semantic-grammatical analysis of the stages (degrees) and limit of adverbial transposition of nouns such as spring, summer, morning, day in the Russian language. The relevance of the study is due to the increased attention in modern linguistics to the problems of interaction between the grammatical and lexical in the structure of linguistic units that undergo various transposition processes, including adverbialization. The object of the study is the linguistic mechanism of adverbialization of nouns, producing adverbs and transitional, syncretic formations that have moved away to varying degrees from the original class of words and have come closer to adverbs; The subject of consideration is the stages (stages, degrees) of functional adverbialization of substantive word forms, reflected in typical contexts from the National Corpus of the Russian Language. The novelty of the approach is associated with the use of the technique of oppositional analysis (and the adverbial transposition scale), aimed at identifying the degrees of adverbialization of nouns. As a result of the study, semantic and grammatical features of core and peripheral nouns, peripheral adverbs and intermediate, hybrid structures were identified; criteria have been established for the differentiation of functional homonyms – non-prepositional forms of the instrumental case of nouns and adverbs formed on their basis.

The Amazon urban weft: a methodological proposal for recognizing a territory of possibilities

In the Amazon, urbanized areas, stretching beyond the cities, constitute a complex urban network, with exchanges of economic and sociocultural capital, mediated by the management of socio-biodiversity. Despite holding the potential for an open future, this territory is generally rendered invisible by the instruments and methods used to understand and represent urbanization in the Amazon. Based on a reinterpretation and analysis of the original classes of the situations of the census tracts, combined with environmental data, this article proposes a methodological alternative that seeks to represent the complex and extensive Amazonian urban weft, considering the state of Pará as the spatial focus and three of the state’s distinct integration regions as the study area. The results have demonstrated that the proposed methodology was able to spatially delineate this weft – much larger than the municipal seats – and highlight areas where the urbanization-nature relationship is still preserved and has more chances to evolve.

Some density-based silhouette diagnostics for soft clustering algorithms

One of the main objectives of cluster analysis is to determine the most effective clustering algorithm. With the wide variety of algorithms available, assessing which one performs better is important. The performance of different clustering methods is typically measured using the Adjusted Rand Index (ARI), which relies on knowledge of the original class labels. However, this study introduces flexible modified alternatives of density-based silhouette methods for evaluating cluster performance. These proposed Density-based silhouettes can be applied to any soft clustering algorithms and do not require the original class labels. Instead, they rely on posterior probabilities. In this study, eight different soft clustering algorithms were evaluated using real and simulated data sets. The goal is to compare their effectiveness and performance using existing and proposed measures based on silhouette and the ARI.

An idea that refuses to die. Rise, fall and resurgence of “housing class”

In 1967, John Rex and Robert Moore published an influential book on ethnic housing inequality in Birmingham, titled Race, Community and Conflict. The book presented a theory about classes formed in the housing market, inspired by a particular reading of Max Weber’s ideas about class. The basic idea of the housing class is that people’s relation to the means of housing is a class-forming factor that is not reducible to their relation to the means of production. This paper presents the original housing class theory and its revision by Peter Saunders, property class theory, and criticisms of the two theories. Since the 1980s, both theories have disappeared from housing research. In the 2020s, the basic idea of the housing class was revived in Lisa Adkins’ and her co-authors’ asset class theory. This paper discusses the history of the idea of the housing class and its contemporary relevance. It also raises questions about the conceptualization of class and housing inequality.

Modular Many-Body Quantum Sensors.

Quantum many-body systems undergoing phase transitions have been proposed as probes enabling beyond-classical enhancement of sensing precision. However, this enhancement is usually limited to a very narrow region around the critical point. Here, we systematically develop a modular approach for introducing multiple phase transitions in a many-body system. This naturally allows us to enlarge the region of quantum-enhanced precision by encompassing the newly created phase boundaries. Our approach is general and can be applied to both symmetry-breaking and topological quantum sensors. In symmetry-breaking sensors, we show that the newly created critical points inherit the original universality class and a simple total magnetization measurement already suffices to locate them. In topological sensors, our modular construction creates multiple bands which leads to a rich phase diagram. In both cases, Heisenberg scaling for Hamiltonian parameter estimation is achieved at all the phase boundaries. This can be exploited to create a global sensor which significantly outperforms a uniform probe.

The latent transition of posttraumatic stress disorder and growth among adolescents surviving Jiuzhaigou earthquake.

Knowledge of longitudinal changes in posttraumatic stress disorder (PTSD) and posttraumatic growth (PTG) may help survivors recover better. Yet, researchers dispute the population-based typologies of PTSD and PTG as well as the temporal transition between these subpopulations across time, especially among adolescents. Thus, the transition pattern still needs further research. Besides, parent-child factors (parenting styles, parent-child cohesion, and parental attachment styles) may influence the transition, but it keeps unclear. In the study, we aimed to solve these questions. A three-wave investigation was conducted among 620 adolescents, 12 months, 21 months, and 27 months, after the Jiuzhaigou earthquake by using self-report measures on PTSD, PTG, and parent-child factors at the three time points. Finally, 339 of them finished the three-wave investigation. Latent profile analysis models showed that three heterogeneous classes of PTSD and PTG existed across time: low-affected (low-level PTSD and PTG), thriving (low-level PTSD but high-level PTG), and struggling (high-level PTSD and PTG) groups. Random intercept latent transition analysis model suggested that samples mainly stayed in the original classes across time, with three main transitional paths: from struggling group to thriving group, from thriving group to low-affected group, and from low-affected group to struggling group. Besides, the study also found that parental rejection, overprotection, and anxious attachment were the possible factors that kept the stability of the struggling group. Parent-child cohesion increased the stability of the thriving group across time. Anxious attachment may worsen PTSD among adolescents and lower the stability of low-affected groups across time. Coexisting and transitional patterns exist in PTSD and PTG across time. Wrong parenting styles and insecure attachments can exacerbate PTSD symptoms and diminish adolescents' resilience, but parent-child cohesion can facilitate their growth after trauma. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Latent transition of social participation and its effects on depression among older adults in China: A National Longitudinal Study

ObjectivesAs populations age, the issue of social participation among older adults has gained prominence. Studies indicate variability in social participation trajectories among this demographic, yet the transition patterns and their effects on depression remain unclear. This longitudinal study aims to explore the latent classes and transition patterns in social participation among older adults and to evaluate their effects on depression. MethodsData from the Chinese Longitudinal Healthy Longevity Survey (CLHLS) in 2014 (T1) and 2018 (T2) were analyzed, including 2293 older adults. Latent class analysis (LCA) and latent transition analysis (LTA) were employed to identify latent classes of social participation at T1 and T2, as well as the transition probabilities between these classes. Multinomial logistic regression was used to examine predictors of transitions, and depression levels at T2 were compared across transition patterns. ResultsThe LCA results supported a 3-class model labeled as low, moderate, and high social participation. The probabilities of remaining stable and transitioning to other classes were similar across the three classes (ranging from 0.50 to 0.54). Age, gender, and other baseline characteristics emerged as significant predictors of transition patterns. Older adults experiencing positive transitions exhibited reduced depression compared to those in their original class over time, while those with negative transitions showed increased depression. ConclusionsThis research prompts a deep understanding of social participation dynamics in older adults and their effects on depression. Identifying social participation classes and transition patterns could inform interventions to enhance social participation and reduce depression among older adults.

Beyond singular prototype: A prototype splitting strategy for few-shot medical image segmentation

In the realm of medical image semantic segmentation, few-shot learning, characterized by its efficient data utilization and flexible generalization capabilities, has been garnering increasing attention. The mainstream methods currently employ prototype-based approaches, which extract semantic knowledge from the annotated support images to guide the segmentation of the query image via masked global average pooling. However, such masked global average pooling leads to severe information loss, which is more problematic for medical images with large numbers of highly heterogeneous background categories. In this work, we propose a prototype splitting module (PSM) to effectively address the issue of semantic information loss in few-shot medical image segmentation. Specifically, PSM iteratively splits the support image masks into set of sub-masks containing segmented regions and unsegmented regions in a self-guided manner. This maximally retains the information within the original semantic classes and better extracts the representations of those classes. Additionally, we devise a multi-level cross attention module (MCAM) that transfers the foreground information from the support images to the query images across different levels to facilitate final segmentation prediction. We validate our method on multiple modal and multi-semantic medical image datasets. Results demonstrate that our approach achieves superior performance over existing state-of-the-art methods. The code has been released on https://github.com/fdngh/PSMnet.

A generation of synthetic samples and artificial outliers via principal component analysis and evaluation of predictive capability in binary classification models

Unbalanced sample groups tend to yield models with a higher prevalence of predominant classes. A sample group with balanced classes contributes to the development of more robust models with improved predictive capability to classify classes equally. In the literature, two methodologies for sample balancing can be found: elimination (undersampling) and synthetic sample generation (oversampling). Undersampling methodologies result in the loss of real samples, while oversampling methods may introduce issues related to adding non-real signals to the original spectra. To overcome these challenges, this paper aimed to utilize Principal Component Analysis (PCA) for the generation of virtual samples (synthetic samples and artificial outliers) to balance data in multivariate classification models. The proposed methodology was applied to data from mid-infrared spectroscopy (MIR) and high-resolution mass spectrometry (HRMS) with Partial Least Squares Discriminant Analysis (PLS-DA) and Support Vector Machine (SVM) models. The constructed models demonstrate that the addition of virtual samples enhances performance parameters (e.g., false negative rate, false positive rate, accuracy, sensitivity, specificity, among others) compared to unbalanced models, while also mitigating overfitting (a problem found in unbalanced models). Performance parameters exhibited a more significant improvement percentage using the non-linear model (SVM) compared to the linear model (PLS-DA). Furthermore, the created virtual spectra do not introduce new signals, i.e., original, and virtual spectra exhibit a similar spectral profile, differing only in the intensity levels. Finally, all models demonstrated good predictive capability according to permutation testing for the binary model developed in this work, limiting the rate of class permutation retention (between 40 % and 60 % of the y-vector remained in the original class). All created models exhibited accuracy values higher than the accuracy distribution of models with permuted classes for the test group.

On the Relaxation of Gauss’s Capillarity Theory Under Spanning Conditions

We study a variational model for soap films in which the films are represented by sets with fixed small volume rather than surfaces. In this problem, a minimizing sequence of completely “wet" films, or sets of finite perimeter spanning a wire frame, may converge to a film containing both wet regions of positive volume and collapsed (dry) surfaces. When collapsing occurs, these limiting objects lie outside the original minimization class and instead are admissible for a relaxed problem. Here we show that the relaxation and the original formulation are equivalent by approximating the collapsed films in the relaxed class by wet films in the original class.

Abstract LB100: Performance of a multi-analyte, multi-cancer early detection (MCED) blood test in a prospectively-collected cohort

Abstract Background: A multi-analyte blood test has potential for robust sensitivity in detecting a broad range of cancer types and stages. Previously, using retrospectively collected samples, we trained and independently assessed in a case-control study the performance of various biomarker classes for the detection of cancers (Douville, Annals of Oncology,2022:pS575). The aim of this study was to further assess two of the four original biomarker classes (methylation and protein) using samples from a large, multi-center, prospectively collected study: Ascertaining Serial Cancer patients to Enable New Diagnostic 2 (ASCEND 2). Methods: The study included 6,314 samples (1,426 cancer and 4,888 non-cancer) collected in LBgard® tubes. The training cohort included 3,026 samples (654 cancers and 2,372 non-cancer controls) and the test set included 3,288 samples (772 cancers and 2,516 non-cancer controls). By measuring biomarkers that capture shared cancer signals, the test is designed to detect a broad range of cancer types. In this analysis, cancer stages were well-represented across 21 solid and hematologic tumor organ sites that collectively represent >85% of incident cancers. We combined DNA methylation and protein biomarkers in a single overarching classifier. Machine learning models were calibrated using stratified cross-validation to reach optimal cancer detection at a combined specificity >99.0%. Results: The specificity threshold identified in training resulted in an overall test set sensitivity of 49.9% and 99.0% specificity, with sensitivities of 23.6% (n=364), 68.6% (n=191), 84.6% (n=182), and 40.0% (n=35) for stages I/II, III, IV, and unknown, respectively. We will also report the cross-validation results of the multi-analyte classifiers in this newly enrolled case-control study. Conclusions: In ASCEND2, a large prospectively collected case-control study that included >85% of cancer burden, new training data and updated models demonstrated robust performance of methylation and protein biomarkers. These results demonstrate reproducibility of the multi-biomarker class approach to cancer detection in the ASCEND 2 cohort and represent an evolution in cancer classifier design. Citation Format: Vladimir Gainullin, Hee Jung Hwang, Larson Hogstrom, Kevin Arvai, Amira Best, Melissa Gray, Madhav Kumar, Mael Manesse, Xi Chen, Philip Uren, Amin Mazloom, Gustavo Cerqueira, Jorge Garces, Tomasz M. Beer, Abigail Mcelhinny, Frank Diehl. Performance of a multi-analyte, multi-cancer early detection (MCED) blood test in a prospectively-collected cohort [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr LB100.

From the History of Leśniewski’s Mereology

Abstract In this paper, we want to present the genesis of Stanisław Leśniewski’s mereology. Although ‘mereology’ comes from theword ‘part’, mereology arose as a theory of collective classes. That is why we present the differences between the concepts of being a distributive class and being a collective class. Next, we present Leśniewski’s original mereology from 1927, but with a modern approach. Leśniewski was inspired to create his concept of classes and their elements by Russell’s antinomy. To face it, Leśniewski had to define the concept of being an element of based on the concept of being part of. Leśniewski showed that in his theory, there is no equivalent to Russell’s antinomy. We will show that his solution has nothing to do with the original approach because, in both cases, we are talking about objects of a different kind. Russell’s original antinomy concerned distributive classes, and Leśsniewski’s considerations concerned collective classes.

Enhanced External Counterpulsation Outcomes Study: Retrospective Analyses of Data Obtained from Patients at a Single Medical Center in United States.

The aim was to explore the effectiveness of enhanced external counterpulsation (EECP) therapy in patients with severe angina pectoris/ chronic heart failure symptoms, who were not suitable candidates for invasive treatment. This retrospective study employed a comprehensive methodology that includes individualized treatment, continuous monitoring, and thorough pre- and postprogram evaluations to assess the efficacy of EECP therapy. The standard protocol involved 35 one-hour treatments, with flexibility for extensions based on therapeutic progress. When pre- and posttreatment results were analyzed, EECP improved the original functional class compared with pretreatment. The mean difference in the functional class was 1.32 (0.92), p < 0.0001. Six-minute walk (6MW) distance improved from 383.6 m (110.24) to 423.1 m (121.50) with mean difference of 37.1 (44.99), p < 0.0001. Duke Activity Status Index (DASI) score improved from 3.9 (2.75) to 6.0 (4.17) with mean difference of 2.16 (3.8), p < 0.0001. Training metabolic equivalents (METs) improved from 3.0 (0.74) to 4.0 (1.57) with mean difference of 1.04 (1.2), p < 0.0001. Weekly anginal events decreased from 13.1 (13.19) to 3.2 (7.38) with mean difference of -9.78 (11.7), p < 0.0001. EECP resulted in improvement of angina pectoris functional class, the 6MW distance, reduction in the number of hospitalizations in first year posttreatment, a significant decrease in sublingual nitroglycerin use, improvement of systolic and diastolic blood pressure, and improvement of DASI score.

Deconfounded multi-organ weakly-supervised semantic segmentation via causal intervention

In weakly-supervised semantic segmentation, obtaining the class activation maps for pseudo masks is crucial. Since multiple organs appear in the same medical image, it is reasonable to obtain the activation maps of each organ by the organ-level features instead of the image-level features. The image-level features are decomposed into the organ-level features, yet the prior anatomical knowledge makes a spurious association between the image-level and organ-level features. To this end, we apply the causal intervention to cut off the spurious association and propose a novel deconfounded multi-organ weakly-supervised semantic segmentation (DeMos) method. Based on the original class activation mapping (CAM) method, the model is retrained to learn the deconfounded features of each organ via cross-attention, and we approximate the expectation of the intervention instead of the traditional likelihood. When the model converges, we extract the activation maps by CAM. Our method not only generates high-quality pseudo masks on the CHAOS, ACDC and ProMRI datasets, but is also applicable to other CAM variants. Furthermore, with the refinement, DeMos achieves the dice similarity coefficient of 93.26% on the task of the left ventricle segmentation, which outperforms the state-of-the-art methods.

Few-shot learning based on prototype rectification with a self-attention mechanism

Few-shot learning (FSL) is a challenging problem. Prototype-based methods are simple and effective methods for addressing few-shot learning. Due to the lack of labeled samples, the learned class prototype in the existing prototype-based few-shot learning methods has a great deviation and cannot express the representative and discriminant characteristics of its corresponding class well. To address this problem, in this work we propose few-shot learning based on prototype rectification with a self-attention mechanism(FSL-PRS). To learn more unbiased and discriminative class prototypes, FSL-PRS takes the support set and the query set as a whole and learns task-related features from the features extracted from pretrained backbone networks with a self-attention mechanism. Then, the learned task-related features are utilized to compute the original class prototypes and predict a pseudo label and confidence for each query sample. The query samples with high confidence are incorporated into the support set to rectify the class prototypes. We hope that the learned class prototype can better highlight the class significance. Therefore, a class significance learning module is designed for making the learned class prototypes more discriminative. Different from prior works, we take the support set and the query set as a whole to learn task-related features with a self-attention mechanism, which not only alleviates the negative effect of distribution differences between the support set and query set but also fuses global context information to enhance features for FSL. We conduct comprehensive experiments on four benchmark datasets widely adopted in few-shot learning. The experimental results demonstrate that the FSL-PRS achieves state-of-the-art performance, which validates its effectiveness.

Discovery of a novel class of reversible monoacylglycerol lipase inhibitors for potential treatment of depression

Biological studies on the endocannabinoid system (ECS) have suggested that monoacylglycerol lipase (MAGL), an essential enzyme responsible for the hydrolysis of 2-arachidonoylglycerol (2-AG), is a novel target for developing antidepressants. A decrease of 2-AG levels in the hippocampus of the brain has been observed in depressive-like models induced by chronic stress. Herein, employing a structure-based approach, we designed and synthesized a new class of (piperazine-1-carbonyl) quinolin-2(1H)-one derivatives as potent, reversible and selective MAGL inhibitors. And detailed structure-activity relationships (SAR) studies were discussed. Compound 27 (IC50 = 10.3 nM) exhibited high bioavailability (92.7%) and 2-AG elevation effect in vivo. Additionally, compound 27 exerted rapid antidepressant effects caused by chronic restraint stress (CRS) and didn't show signs of addictive properties in the conditioned place preference (CPP) assays. Our study is the first to report that reversible MAGL inhibitors can treat chronic stress-induced depression effectively, which may provide a new potential therapeutic strategy for the discovery of an original class of safe, rapid antidepressant drugs.

Belonging in Surgery: A Validated Instrument and Single Institutional Pilot.

The aim of this study was to develop and validate an instrument to measure Belonging in Surgery among surgical residents. Belonging is the essential human need to maintain meaningful relationships and connections to one's community. Increased belongingness is associated with better well-being, job performance, and motivation to learn. However, no tools exist to measure belonging among surgical trainees. A panel of experts adapted a belonging instrument for use among United States surgery residents. After administration of the 28-item instrument to residents at a single institution, a Cronbach alpha was calculated to measure internal consistency, and exploratory principal component analyses were performed. Multiple iterations of analyses with successively smaller item samples suggested the instrument could be shortened. The expert panel was reconvened to shorten the instrument. Descriptive statistics measured demographic factors associated with Belonging in Surgery. The overall response rate was 52% (114 responses). The Cronbach alpha among the 28 items was 0.94 (95% CI: 0.93-0.96). The exploratory principal component analyses and subsequent Promax rotation yielded 1 dominant component with an eigenvalue of 12.84 (70% of the variance). The expert panel narrowed the final instrument to 11 items with an overall Cronbach alpha of 0.90 (95% CI: 0.86, 0.92). Belonging in Surgery was significantly associated with race (Black and Asian residents scoring lower than White residents), graduating with one's original intern cohort (residents who graduated with their original class scoring higher than those that did not), and inversely correlated with resident stress level. An instrument to measure Belonging in Surgery was validated among surgical residents. With this instrument, Belonging in Surgery becomes a construct that may be used to investigate surgeon performance and well-being.

Synthesized Image Training Techniques: On Improving Model Performance using Confusion.

The performance of supervised deep learning image classifiers has significantly improved with large, labeled datasets and increased computing power. However, obtaining large, labeled image datasets in areas like medicine is expensive. This study seeks to improve model performance on limited labeled datasets by reducing confusion. We observed that misclassification (or confusion) between classes is usually more prevalent between specific classes. Thus, we developed synthesized image training techniques (SIT2), a novel confusion-based training framework that identifies pairs of classes with high confusion and synthesizes not-sure images from these pairs. The not-sure images are utilized in three new training strategies as follows. (1) The not-sure training strategy pretrains a model using not-sure images and the original training images. (2) The sure-or-not strategy pretrains with synthesized sure or not-sure images. (3) The multi-label strategy pretrains with synthesized images but predicts the original class(es) of the synthesized images. Finally, the pretrained model is finetuned on the original dataset. An extensive evaluation was conducted on five medical and non-medical datasets. Several improvements are statistically significant, which shows the promising future of our confusion-based training framework.

ISODF-ENN:Imbalanced data mixed sampling method based on improved diffusion model and ENN

In the era of big data, the complexity of data is increasing. Problems such as data imbalance and class overlap pose challenges to traditional classifiers. Meanwhile, the importance of imbalanced data has become increasingly prominent, it is necessary to find appropriate methods to enhance classification performance of classifiers on such datasets. In response, this paper proposes a mixed sampling method (ISODF-ENN) based on iterative self-organizing (ISODATA) denoising diffusion algorithm and edited nearest neighbors (ENN) data cleaning algorithm. The algorithm first uses iterative self-organizing clustering algorithm to divide minority class into different sub-clusters, then it uses denoising diffusion algorithm to generate new minority class data for each sub-cluster, and finally it uses ENN algorithm to preprocess majority class data to remove the overlap with the minority class data. Each sub-cluster is oversampled according to sampling ratio, so that the oversampled minority class data also conforms to the distribution of original minority class data. Experimental results on keel datasets demonstrate that the proposed method outperforms other methods in terms of F-value and AUC, effectively addressing the issues of class imbalance and class overlap.