Hierarchical Classification Research Articles

On the complexity of computing the “Shuffled Inequality” function in classical and quantum NOBDDs

We investigate Ordered Binary Decision Diagrams (OBDDs) — a model for computing Boolean functions. It is known that OBDD’s complexity can extremely depend on the order of reading variables. There are techniques for constructing functions that do not allow choosing the optimal order for reading the input, one of which we use in this paper. A function “Shuffled Inequality” NEQS is presented, for which a lower and an upper bounds for the complexity of nondeterministic OBDD are proved. The upper bound is an improvement of a previously known result. A quantum measure-many non-deterministic OBDD is constructed that is more efficient than the classical one. The hierarchy of complexity classes defined on the basis of OBDD models is clarified.

Enhancing aboveground biomass estimation in Moso bamboo forests: the role of on-year and off-year phenomena in remote sensing

Accurate estimation of aboveground biomass (AGB) in Moso bamboo forests (MBFs) has garnered significant attention over the past two decades. However, the remote sensing-based estimation of AGB in MBFs remains challenging because of the limited understanding of the relationship between Moso bamboo growth characteristics and remote sensing data, particularly concerning alternating on-year and off-year cycles. In this study, Sentinel-2 remote sensing imagery and plot survey data were selected, a novel change detection algorithm to assess plot level AGB dynamics between 2018 and 2019 was developed, a hierarchical classifier was proposed to map the spatial distributions of on-year and off-year MBFs, and a time series model was developed for estimating the AGB of MBFs to characterize AGB dynamics between November and December. The results indicated that the AGB of the MBFs exhibited a distinct dynamic cycle characterized by the rapid accumulation of new bamboo and sharp reductions due to selective harvesting during the on-year period, alongside a steady accumulation of lignified bamboo during the off-year period. The AGB of the MBFs during the on-year and off-year cycles ranged primarily from 30 to 80 Mg/ha, with the AGB of the on-year MBFs generally exceeding that of the off-year MBFs. This study demonstrated the potential to accurately estimate AGB and its dynamic changes by accounting for on-year and off-year phenomena.

Few-shot Hierarchical Text Classification with Bidirectional Path Constraint by label weighting

Few-shot Hierarchical Text Classification with Bidirectional Path Constraint by label weighting

Hierarchical image classification using transfer learning to improve deep learning model performance for amazon parrots

Numerous studies have proven the potential of deep learning models for classifying wildlife. Such models can reduce the workload of experts by automating species classification to monitor wild populations and global trade. Although deep learning models typically perform better with more input data, the available wildlife data are ordinarily limited, specifically for rare or endangered species. Recently, citizen science programs have helped accumulate valuable wildlife data, but such data is still not enough to achieve the best performance of deep learning models compared to benchmark datasets. Recent studies have applied the hierarchical classification of a given wildlife dataset to improve model performance and classification accuracy. This study applied hierarchical classification by transfer learning for classifying Amazon parrot species. Specifically, a hierarchy was built based on diagnostic morphological features. Upon evaluating model performance, the hierarchical model outperformed the non-hierarchical model in detecting and classifying Amazon parrots. Notably, the hierarchical model achieved the mean Average Precision (mAP) of 0.944, surpassing the mAP of 0.908 achieved by the non-hierarchical model. Moreover, the hierarchical model improved classification accuracy between morphologically similar species. The outcomes of this study may facilitate the monitoring of wild populations and the global trade of Amazon parrots for conservation purposes.

CPM-based Hierarchical Text Classification

In the field of natural language processing, hierarchical text classification (HTC) has emerged as a critical task for organizing and analyzing large volumes of text data. The previous work of HTC often falls short in fully leveraging the hierarchical structure of labels, resulting in suboptimal performance. In addition, it is difficult to capture nuanced relationships between parent and child classes, leading to inaccurate predictions and insufficient differentiation between sibling classes under the same parent category. This gap underscores the need for approaches that can more effectively integrate and utilize both hierarchical and corpus-specific information to improve HTC performance. To address these issues, Concept-aware Prompt Mechanism (CPM) is proposed for HTC, which leverages concept information embedded within hierarchical labels to enhance the representation of these labels and improve classification accuracy. Specifically, we introduce a concept initialization module that extracts concept features from hierarchical labels and a novel concept prompt template to integrate these features into the classification process. Our experimental results demonstrate that the proposed CPM achieves state-of-the-art performance on two benchmark datasets, improving Micro-F1 and Macro-F1 scores to varying degrees, particularly in datasets with complex label hierarchies.

Hierarchical Multi-Label Classification With Gene-Environment Interactions in Disease Modeling.

In biomedical studies, gene-environment (G-E) interactions have been demonstrated to have important implications for analyzing disease outcomes beyond the main G and main E effects. Many approaches have been developed for G-E interaction analysis, yielding important findings. However, hierarchical multi-label classification, which provides insightful information on disease outcomes, remains unexplored in G-E analysis literature. Moreover, unlabeled data are commonly observed in practical settings but omitted by many existing methods of hierarchical multi-label classification. In this study, we consider a semi-supervised scenario and develop a novel approach for the two-layer hierarchical response with G-E interactions. A two-step penalized estimation is then proposed using an efficient expectation-maximization (EM) algorithm. Simulation shows that it has superior performance in classification and feature selection. The analysis of The Cancer Genome Atlas (TCGA) data on lung cancer demonstrates the practical utility of the proposed method. Overall, this study can fill the important knowledge gap in G-E interaction analysis by providing a widely applicable framework for hierarchical multi-label classification of complex disease outcomes.

SPIRIT: Structural Entropy Guided Prefix Tuning for Hierarchical Text Classification

Hierarchical text classification (HTC) is a challenging task that requires classifiers to solve a series of multi-label subtasks considering hierarchical dependencies among labels. Recent studies have introduced prompt tuning to create closer connections between the language model (LM) and the complex label hierarchy. However, we find that the model’s attention to the prompt gradually decreases as the prompt moves from the input to the output layer, revealing the limitations of previous prompt tuning methods for HTC. Given the success of prefix tuning-based studies in natural language understanding tasks, we introduce Structural entroPy guIded pRefIx Tuning (SPIRIT). Specifically, we extract the essential structure of the label hierarchy via structural entropy minimization and decode the abstractive structural information as the prefix to prompt all intermediate layers in the LM. Additionally, a depth-wise reparameterization strategy is developed to enhance optimization and propagate the prefix throughout the LM layers. Extensive evaluation on four popular datasets demonstrates that SPIRIT achieves a state-of-the-art performance.

Medicare Risk Arrangement and Use and Outcomes Among Physician Groups

Many physician groups are in 2-sided risk payment arrangements with Medicare Advantage plans (at-risk MA). Analysis of quality and health resource use under such arrangements may inform ongoing Medicare policy concerning payment and service delivery. To compare quality and efficiency measures under 2 payment models: at-risk MA and fee-for-service (FFS) MA. This cross-sectional study used Medicare encounter and enrollment data from 2016 to 2019 covering 17 physician groups, 15 488 physicians, and 35 health insurers to compare quality and health resource use for Medicare beneficiaries within the same physician groups. The data were analyzed between August 4 and October 30, 2024. Care delivered under at-risk MA and FFS MA payment arrangements by the same physicians and medical groups. Twenty quality and efficiency measures across 4 domains of patient care (hospital care, avoidance of the emergency department [ED], avoidance of disease-specific admissions, and outpatient care) were examined using logistic regression analysis. The overall sample comprised 5 278 717 person-years (37.7% at-risk MA and 62.3% FFS MA). The mean (SD) age of beneficiaries was 73.6 (9.2) years in the at-risk MA group (56.8% women) and 71.8 (10.4) years in the FFS MA group (57.4% women). For at-risk MA compared with FFS MA, inpatient admissions and 30-day readmissions per 1000 were 10.03 (95% CI, -10.61 to -9.44) and 1.95 (95% CI, -2.18 to -1.73) lower. ED use measures per 1000 ranged from 2.95 (95% CI, -3.28 to -2.63) lower for avoidable ED visits to 26.02 (95% CI, -26.92 to -25.12) lower for overall ED visits. Avoidance of disease-specific admissions per 1000 ranged from 0.24 (95% CI, -0.35 to -0.13) lower for composite diabetes-related admissions to 2.18 (95% CI, -2.43 to -1.94) lower for the composite of chronic disease-related admissions. High-risk drug use per 1000 was 14.26 (95% CI, -14.85 to -13.67) lower. Overall, compared with FFS MA, at-risk MA was associated with higher quality and efficiency in 18 of 20 measures after adjusting for differences in demographics, Hierarchical Condition Categories Risk Adjustment Factor scores, and other health characteristics. In this cross-sectional study, at-risk MA payment arrangements managed by physician groups were associated with higher quality and efficiency compared with FFS MA managed by the same groups. The population and methods used provide robust evidence that at-risk payment arrangements in MA may improve health care delivery for the MA population.

A dataset of Antarctic ecosystems in ice-free lands: classification, descriptions, and maps

Antarctica, Earth’s least understood and most remote continent, is threatened by human disturbances and climate-related changes, underscoring the imperative for biodiversity inventories to inform conservation. Antarctic ecosystems support unique species and genetic diversity, deliver essential ecosystem services and contribute to planetary stability. We present Antarctica’s first comprehensive ecosystem classification and map of ice-free lands, which host most of the continent’s biodiversity. We used latent variables in factor analyses to partition continental-scale abiotic variation, then biotic variation represented in spatial models, and finally recognised regional-scale variation among biogeographic units. This produced a spatially explicit hierarchical classification with nine Major Environment Units (Tier 1), 33 Habitat Complexes (Tier 2) and 269 Bioregional Ecosystem Types (Tier 3) mapped at 100 m resolution and aligned with ‘level 4’ of the IUCN Global Ecosystem Typology. This comprehensive ecosystem inventory provides foundational data to inform protected area designation under the Antarctic Treaty’s Environmental Protocol and track risks to Antarctic ecosystems. Its tiered structure and workflow accommodate data scarcity and facilitate updates, promoting robustness as knowledge builds.

Clustering-Based Class Hierarchy Modeling for Semantic Segmentation Using Remotely Sensed Imagery

Land use/land cover (LULC) nomenclature is commonly organized as a tree-like hierarchy, contributing to hierarchical LULC mapping. The hierarchical structure is typically defined by considering natural characteristics or human activities, which may not optimally align with the discriminative features and class relationships present in remotely sensed imagery. This paper explores a novel cluster-based class hierarchy modeling framework that generates data-driven hierarchical structures for LULC semantic segmentation. First, we perform spectral clustering on confusion matrices generated by a flat model, and then we introduce a hierarchical cluster validity index to obtain the optimal number of clusters to generate initial class hierarchies. We further employ ensemble clustering techniques to yield a refined final class hierarchy. Finally, we conduct comparative experiments on three benchmark datasets. Results demonstrating that the proposed method outperforms predefined hierarchies in both hierarchical LULC segmentation and classification.

Towards Detecting Associations of Canine Astrovirus and Caliciviruses with Health and Living Characteristics of Dogs in Greece.

Astroviruses and caliciviruses are important causative agents of gastroenteritis in humans worldwide. They have been detected in a variety of animal species, including dogs, but their role in the induction of disease in animals remains uncertain. In a molecular study that was conducted in Greece, including healthy and gastroenteritis-affected dogs of different ages, astrovirus (AstV) and sapovirus (SaV) were detected in 15% and 26% of the examined animals, respectively. A specialized questionnaire was filled out for each of the dogs participating in the study, including information about different characteristics and risk factors that could possibly affect their health status. This information was analyzed with the use of two innovative statistical methods, i.e., a Multiple Correspondence Analysis (MCA) and the Ascending Hierarchical Classification (AHC). Based on their results, it was possible to define various groups of dogs based on their characteristics. AstV seems to occur more often in low-health-status dogs, usually mongrels, living in rural areas, showing vomit, diarrhea, and diet changes. Dogs of this group usually live with other pets in the same household and have frequent contact with stray animals. The presence of SaV does not seem to be associated with any of the examined factors.

Morphological Classification of Galaxies Through Structural and Star Formation Parameters Using Machine Learning

Abstract We employ the XGBoost machine learning (ML) method for the morphological classification of galaxies into two (early-type, late-type) and five (E, S0–S0a, Sa–Sb, Sbc–Scd, Sd–Irr) classes, using a combination of non-parametric (C, A, S, AS, Gini, M20, c5090), parametric (Sérsic index, n), geometric (axial ratio, BA), global colour (g − i, u − r, u − i), colour gradient (Δ(g − i)), and asymmetry gradient (ΔA9050) information, all estimated for a local galaxy sample (z < 0.15) compiled from the Sloan Digital Sky Survey (SDSS) imaging data. We train the XGBoost model and evaluate its performance through multiple standard metrics. Our findings reveal better performance when utilizing all fourteen parameters, achieving accuracies of 88% and 65% for the two-class and five-class classification tasks, respectively. In addition, we investigate a hierarchical classification approach for the five-class scenario, combining three XGBoost classifiers. We observe comparable performance to the “direct” five-class classification, with discrepancies of only up to 3%. Using SHAP (an advanced interpretation tool), we analyse how galaxy parameters impact the model’s classifications, providing valuable insights into the influence of these features on classification outcomes. Finally, we compare our results with previous studies and find them consistently aligned.

Hierarchical skin lesion image classification with prototypical decision tree

Traditional disease classification models often disregard the clinical significance of misclassifications and lack interpretability. To overcome these challenges, we propose a hierarchical prototypical decision tree (HPDT) for skin lesion classification. HPDT combines prototypical networks and decision trees, leveraging a class hierarchy to guide interpretable predictions from general to specific categories. By incorporating a hierarchy-based distance matrix, the model prioritizes less severe misclassifications while maintaining diagnostic accuracy. Evaluated on a dataset of 235,268 dermoscopic images across 65 conditions, HPDT outperforms flat classifiers and existing hierarchical methods in accuracy, error severity reduction, and interpretability. It also generalizes effectively to unseen classes. These results highlight the value of integrating clinical hierarchies into model design and training to improve diagnostic reliability and decision transparency, demonstrating HPDT’s potential for clinical decision support.

Indonesian Tourism Workers on Volcanoes and Geotourism's Colonial Origins: Making a Subaltern History Visible

Abstract Indonesia's volcanoes are places of recreation, aesthetic production, and scientific knowledge-gathering, as well as sites of pilgrimage, spirituality, and natural disasters for locals as well as international travellers. In this article, I focus on volcanoes as historic sites of labour to demonstrate the entanglement of colonial tourism and science with local forms of work and knowledge, and to reveal the origins of the porting and guiding work that takes place on Indonesia's volcanoes to this day. Using Tina Campt's method of “listening to images,” I show how colonial photographs, albeit partial sources, make modes of subaltern labour visible that written sources routinely minimised, restoring porters, guides, and what I call “camp domestics” to histories of service, science and geotourism in Indonesia. Recognising the homosocial setting of the colonial scientific expedition and the peculiar physical challenges of the volcano environment, I also examine the negotiation of Indonesian and European masculinities and their intersection with class and racial hierarchies on the volcano. The article thus reflects on how Javanese workers’ spatial and social mobility entailed the negotiation of opportunity as well as exploitation on tour.

“Come to Java”: Colonial Tourism and the Fragile Illusion of an “Island Paradise”

Abstract The proliferation of mountain resorts and the establishment of tourist itineraries in late colonial Indonesia reflected and reaffirmed colonial hierarchies of race, class, and gender. In the mild mountain climate, Europeans visited hotels reminiscent of European architectural styles, consumed European meals and enjoyed flora and fauna that reminded them of home. The location of these resorts was inspired by nearby natural, cultural, and ethnographic “wonders” that further served to emphasise colonial superiority. As spaces where pseudo-scientific concerns over race, civilisation, and the environment coalesced to legitimise colonial ideology, these mountain resorts were crucial elements in the development of a regional and national tourist network. The infrastructure needed to access these resorts and the identification of tourist destinations in the colonial era laid the foundation for many of Indonesia's contemporary tourist draws. Furthermore, thanks to a targeted international advertising campaign, these sites drew visitors both from within the Netherlands Indies as well from throughout colonial Asia and Oceania. Finally, these tourist networks traversed colonial (national) boundaries and played an important role in the articulation and experience of a shared colonial mindset. This history illustrates the extent to which contemporary tourism is profoundly shaped by the colonial discourse of power and difference.

A Joint Network of 3D-2D CNN Feature Hierarchy and Pyramidal Residual Model for Hyperspectral Image Classification

A Joint Network of 3D-2D CNN Feature Hierarchy and Pyramidal Residual Model for Hyperspectral Image Classification

Hierarchical classification of dynamic carbon emission factors based on improved support vector machine

Hierarchical classification of dynamic carbon emission factors based on improved support vector machine

Hierarchical Method for Automated Text Documents Classification

Digitalization is currently not a concept the world seeks to apply; rather, it is a fact this world lives in. The transformation for the green world has strongly introduced the principle of eliminating hard copy resources while maintaining their digital versions. The immense amount of information that resides in electronic documents opened a wide road for research a long time ago. On the other hand, information extraction, text mining, and Natural Language Processing (NLP) are three concatenated fields that have gained their unique place in the digital world through time. This research aims to introduce a novel method for Arabic document classification. The research provides multi-tagging to the document according to a set of criteria, one of these tags is the hierarchical classification for the document that could play an efficient role in its related field. For example, documents in healthcare systems beehive could lead to exploring a new symptom of a disease, as it is known that symptoms could continuously mutate over time. The proposed method succeeds through the generated schema to relate between old and new symptoms, which makes it no surprise when evolving and gives a chance for pre-preparation and success to containment. The technical challenges of this study include the ability to successfully apply text mining techniques and machine learning. Additionally, the higher level of challenges that arise in this study is the fact that the processing is applied to Arabic text documents. Arabic has been known to be a complex language as it has its unique nature. The proposed method has been applied, compared with known methods, and its effectiveness has been confirmed by applying a classification task with an Accuracy equal to 99.5%

Multimodal hierarchical classification of CITE-seq data delineates immune cell states across lineages and tissues.

Single-cell RNA sequencing (scRNA-seq) is invaluable for profiling cellular heterogeneity and transcriptional states, but transcriptomic profiles do not always delineate subsets defined by surface proteins. Cellular indexing of transcriptomes and epitopes (CITE-seq) enables simultaneous profiling of single-cell transcriptomes and surface proteomes; however, accurate cell-type annotation requires a classifier that integrates multimodal data. Here, we describe multimodal classifier hierarchy (MMoCHi), a marker-based approach for accurate cell-type classification across multiple single-cell modalities that does not rely on reference atlases. We benchmark MMoCHi using sorted T lymphocyte subsets and annotate a cross-tissue human immune cell dataset. MMoCHi outperforms leading transcriptome-based classifiers and multimodal unsupervised clustering in its ability to identify immune cell subsets that are not readily resolved and to reveal subset markers. MMoCHi is designed for adaptability and can integrate annotation of cell types and developmental states across diverse lineages, samples, or modalities.

Reproducible and Generalizable Framework for Multi-class Hierarchical Classification Demonstrated by DNA Methylation-based Glioma Subtyping

Reproducible and Generalizable Framework for Multi-class Hierarchical Classification Demonstrated by DNA Methylation-based Glioma Subtyping