Search

The scarcity of labeled data is a common problem in supervised classification and in particular in hierarchical classification. Therefore, in this work a semi-supervised hierarchical classifier based on local information (SSHC-BLI) is proposed in order to take advantage of labeled and unlabeled data to perform classification tasks. SSHC-BLI is a semi-supervised learning algorithm for hierarchical classification, which tries to pseudo-label each unlabeled instance using the labels of its labeled neighbors, also, it uses a similarity function to determine whether the unlabeled instance is similar to its labeled neighbors to be pseudo-labeled; in this way, the heuristic function similarity of an instance with a set of instances is proposed. SSHC-BLI was tested in several datasets from different fields, including: artificial, functional genomics and text; also, it was compared against a supervised hierarchical classifier and two state of the art methods, showing in most cases superior performance with statistical significance in exact match and Matthews correlation coefficient. [ABSTRACT FROM AUTHOR]

Human Activity Recognition (HAR) plays a crucial role in computer vision and signal processing, with extensive applications in domains such as security, surveillance, and healthcare. Traditional machine learning (ML) approaches for HAR have achieved commendable success but face limitations such as reliance on handcrafted features, sensitivity to noise, and challenges in handling complex temporal dependencies. These limitations have spurred interest in deep learning (DL) and hybrid models that address these issues more effectively. Thus, DL has emerged as a powerful approach for HAR, surpassing the performance of traditional methods. In this paper, a multi-modal hybrid hierarchical classification approach is proposed. It combines DL transformers with traditional ML techniques to improve both the accuracy and efficiency of HAR. The proposed classifier is evaluated based on the different activities of four widely used benchmark datasets: PAMAP2, CASAS, UCI HAR, and UCI HAPT. The experimental results demonstrate that the hybrid hierarchical classifier achieves remarkable accuracy rates of 99.69%, 97.4%, 98.7%, and 98.6%, respectively, outperforming traditional classification methods and significantly reducing training time compared to sequential LSTM models. [ABSTRACT FROM AUTHOR]

One of the key factors in manufacturing products by fused deposition molding (FDM) or layer-by-layer printing technology is the material intensity of the product. The task of reducing the amount of material required to manufacture a product without significant loss of mechanical properties is one of the most practically important technological tasks. Material saving in FDM printing of products allows to reduce financial costs and increase the speed of manufacturing of the final product without reducing (or not significantly reducing) the quality properties of the product. In our work it is demonstrated that using Combs filling type and materials of poly lactic acid (PLA) and polyethylene terephthalate glycol (PETG) it is possible to achieve material savings of up to 23% at 50% filling for PLA and 17% at 75% filling for PETG without significant reduction of product strength in comparison with other filling types. Exceptions are PLA samples with 100% fill and Lateral fill. Application of Kruskal-Wallis criterion and Dunn's criterion with Bonferroni multiple comparison correction showed that there were no statistically significant differences within the strength limits of samples made by FDM printing technology from PLA and PETG plastics (p-value = 0.0514), as well as samples with Triangle and Grid filling type (p-value = 1). Based on this result, three groups of samples statistically significantly differing in ultimate strength were identified by methods of hierarchical cluster analysis; in each group (except for group 1, which included samples made of PLA plastic with Lateral filling type and 100% filling), correlation analysis was performed (Spearman correlation was used). The results of the correlation analysis showed a stable average correlation between the percentage of filling, modulus along the secant 0.05–0.2% strain, ultimate strength and strain corresponding to the yield stress. Analysis of the correlation graph showed that the main parameter correlating with all mechanical properties of the specimen is the 0.05–0.2% strain modulus. Based on this conclusion, robust regression equations predicting the 0.05–0.2% strain modulus as a function of the percentage of specimen filling were constructed for the two selected groups. Analysis of the equations showed that in the third group of specimens, the average modulus of 0.05–0.2% strain is more than twice the modulus of 0.05–0.2% strain in the second group. The detected statistical regularities can be explained by the mechanism of strain hardening, the actual value of which depends on the structure of the macrodefect (type of filling), properties and volume of the material (percentage of filling) used in the fabrication of samples using FDM printing technology. [ABSTRACT FROM AUTHOR]

Copyright of Industrial Minerals & Processing / Huagong Kuangwu yu Jiagong is the property of Industrial Minerals & Processing Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Few-shot learning has become an important branch of machine learning, which aims to give correct prediction information to unknown samples. Many few-shot models mostly adopt a hierarchical structure and make good use of the sample information at different levels of granularity to finish class prediction. However, existing few-shot learning only focuses on the amount of information correctly predicted and ignores the different risk levels of an incorrect prediction. For example, in the field of medical image pathology analysis, difficult pathological images can only be roughly distinguished as whether they are diseased and cannot be specifically analyzed for the degree of disease. In this paper, we propose a hierarchical few-shot learning based on top-down correction mechanism with stop strategy (HTDS) for few-shot classification. Firstly, we get the coarse-grained of each class by taking the mean value for each class, and we use the coarse-grained and the original fine-grained samples to build a decision tree with a multi-granularity hierarchical structure. The decision tree represents the class relationship among samples. Secondly, a threshold is established to measure the impact of coarse-grained on test sample classification. It stops the correct classification at the coarse-grained level and transfers the incorrect classification down to the fine-grained level. Finally, the fine-grained layer assists the coarse-grained layer in classification via a top-down correction mechanism. It minimizes the coarse-grained risk of misclassification. By using stop strategy and revision mechanisms, we can differentiate pathological images and use auxiliary information to assist in different levels of classification in the field of medical image analysis. The experimental results show that our model is better than some popular multi-granularity hierarchical based few-shot learning models. [ABSTRACT FROM AUTHOR]

In the universe, there are up to 2 trillion galaxies with different features ranging from the number of stars, light spectrum, age, or visual appearance. Consequently, automatic classifiers are required to perform this task; furthermore, as shown by some related works, while greater the number of classes considered, the performance of the classifiers tends to decrease. This work is focused on the morphological classification of galaxies. They can be associated with a subset of 10 classes arranged in a hierarchy derived from the Hubble sequence. The proposed method, Bayesian and Convolutional Neural Networks (BCNN), is composed of two main modules. The first module is a convolutional neural network trained with the images of galaxies, and its predictions feed the second module. The second module is a Bayesian network that evaluates the hierarchy and helps to improve the prediction accuracy by combining the predictions of the first module through probabilistic inference over the Bayesian network. A collection of galaxies sourced from the Principal Galaxies Catalog and the APM Equatorial Catalogue of Galaxies are used to perform the experiments. The results show that BCNN performed better than five CNNs in multiple evaluation measures, reaching the scores 83% in hierarchical F-measure, 78% in accuracy, and 67% in exact match evaluation. [ABSTRACT FROM AUTHOR]

Industry assignment, which assigns firms to industries according to a predefined industry classification system (ICS), is fundamental to a large number of critical business practices, ranging from operations and strategic decision-making by firms to economic analyses by government agencies. Three types of expert knowledge are essential to effective industry assignment: definition-based knowledge (i.e., expert definitions of each industry), structure-based knowledge (i.e., structural relationships among industries as specified in an ICS), and assignment-based knowledge (i.e., prior firm-industry assignments performed by domain experts). Existing industry assignment methods utilize only assignment-based knowledge to learn a model that classifies unassigned firms to industries, overlooking definition-based and structurebased knowledge. Moreover, these methods only consider which industry a firm has been assigned to, ignoring the time-specificity of assignment-based knowledge, i.e., when the assignment occurs. To address the limitations of existing methods, we propose a novel deep learning-based method that not only seamlessly integrates the three types of knowledge for industry assignment but also takes the timespecificity of assignment-based knowledge into account. Methodologically, our method features two innovations: dynamic industry representation and hierarchical assignment. The former represents an industry as a sequence of time-specific vectors by integrating the three types of knowledge through our proposed temporal and spatial aggregation mechanisms. The latter takes industry and firm representations as inputs, computes the probability of assigning a firm to different industries, and assigns the firm to the industry with the highest probability. We conduct extensive evaluations with two widely used ICSs and demonstrate the superiority of our method over prevalent existing methods. [ABSTRACT FROM AUTHOR]

With the development of cross-border telemedicine, international cooperation in clinical trials of drugs and vaccines, cross-border academic discussions and clinical research, cross-border flow of health and medical data has become an irreversible trend. Rules for cross-border flow of health data require doing a balance among the objectives and policies of maintaining national health security, protecting citizens' privacy, and promoting the development of health industry. This paper uses literature research, comparative analysis and other methods to study the legislative regulation and regulatory mechanism of cross-border health data in Australia. The study's findings reveal that Australia prioritizes health data security, enforcing stringent oversight on the cross-border movement of personally identifiable health data through dedicated agencies and comprehensive legislation. China should build a scientific, complete and strict legal and policy system for health data protection and cross-border flow, and organically combine and implement laws and regulations with policy guidelines; improve the terms of standard contract for health data exit and play the function of the standard contract for regulation; actively use the existing hierarchical classification of health data to build a flexible and diverse health data exit regulation model; and strengthen exchanges and cooperation with countries along the "Belt and Road", especially ASEAN member countries, in the field of cross-border health data flow.

Purpose: analysis and construction of hierarchical and facet classifications of reclamation pumping stations (RPS). During the research, a search, analysis, comparison and systematization of RPS types were carried out in accordance with regulatory and technical documentation, as well as the developments of various Russian scientists. When developing and constructing RPS classifications, the following methods were used: facet and hierarchical. Discussion. As a result of the study and analysis of various sources of information, it was found that the existing RPS classifications in most cases are presented according to one classification criterion. In this regard, a classification has been carried out at several levels. First, the main classification criteria such as the purpose and main characteristics, including flow and pressure, were determined, according to which the hierarchical classification was built. Subsequently, lower order classification features were considered: importance classes, degree of durability, design, type of building, structure configuration, location relative to the irrigation source, location in the system, degree of use, type of water intake structure and source, devices used and control principle. Based on the indicated classification criteria, hierarchical and facet classifications of the RPS were formed. Conclusions. As a result of research, a hierarchical classification of the RPS by main types with a hierarchy of vague subordination up to the seventh level with minor groups of classification characteristics has been developed. The use of the facet method made it possible to combine all types and kinds of RPS into a more detailed classification with the introduction of some independent classification features of lower levels, specifying certain types of RPS within the hierarchical classification levels. The developed hierarchical and facet classifications of RPS make it possible to their further use in the development of a regulatory document on RPS design.

Temporal distributions of Antarctic krill (Euphausia superba) density and aggregation types were characterized and compared using Nortek Signature100 and SIMRAD Wideband Autonomous Transceiver (WBAT) upward-looking echosounders. Noise varied between the two echosounders. With the Signature100, it was necessary to correct data for background, transient, and impulse noises, while the WBAT data needed to be corrected for background noise only. For selected regions with no visible backscatter, the signal-to-noise ratio of S v values (i.e. the ratio between the signal and the background noise level) did not vary between the two echosounders. Surface echo backscatter was similar during similar time periods. Descriptive metrics were used to quantify spatial and temporal krill vertical distributions: volume backscatter, mean depth, center of mass, inertia, equivalent area, aggregation index, and proportion occupied. Krill backscatter density differed between the two instruments but was detected at similar mean depths. Krill aggregations were identified at each mooring location and classified in three types based on morphological characteristics. Each type of aggregation shape differed at the two spatially separated moorings, while the acoustic density of each aggregation type was similar. The Signature100 detected a lower number of krill aggregations (n  = 133) compared to the WBAT (n  = 707). Although both instruments can be used for autonomous deployment and sampling of krill over extended periods, there is a strong caveat for the use of the Signature100 due to significant differences in noise characteristics and krill detection. [ABSTRACT FROM AUTHOR]

In the analysis of qualification stage data from FIRST Robotics Competition (FRC) championships, the ratio (1.67–1.68) of the number of observations (110–114 matches) to the number of parameters (66–68 robots) in each division has been found to be quite small for the most commonly used winning margin power rating (WMPR) model. This usually leads to imprecise estimates and inaccurate predictions in such three-on-three matches that FRC tournaments are composed of. With the recognition of a clustering feature in estimated robot strengths, a more flexible model with latent clusters of robots was proposed to alleviate overparameterization of the WMPR model. Since its structure can be regarded as a dimension reduction of the parameter space in the WMPR model, the identification of clusters of robot strengths is naturally transformed into a model selection problem. Instead of comparing a huge number of competing models (7.76 × 10 67 to 3.66 × 10 70) , we develop an effective method to estimate the number of clusters, clusters of robots and robot strengths in the format of qualification stage data from the FRC championships. The new method consists of two parts: (i) a combination of hierarchical and non-hierarchical classifications to determine candidate models; and (ii) variant goodness-of-fit criteria to select optimal models. In contrast to existing hierarchical classification, each step of our proposed non-hierarchical classification is based on estimated robot strengths from a candidate model in the preceding non-hierarchical classification step. A great advantage of the proposed methodology is its ability to consider the possibility of reassigning robots to other clusters. To reduce overestimation of the number of clusters by the mean squared prediction error criteria, corresponding Bayesian information criteria are further established as alternatives for model selection. With a coherent assembly of these essential elements, a systematic procedure is presented to perform the estimation of parameters. In addition, we propose two indices to measure the nested relation between clusters from any two models and monotonic association between robot strengths from any two models. Data from the 2018 and 2019 FRC championships and a simulation study are also used to illustrate the applicability and superiority of our proposed methodology. [ABSTRACT FROM AUTHOR]

In the era of big data, there exist complex structure between different classes labels. Hierarchical structure, among others, has become a representative one, which is mathematically depicted as a tree-like structure or directed acyclic graph. Most studies in the literature focus on static feature selection in hierarchical information system. In this study, in order to solve the incremental feature selection problem of hierarchical classification in a dynamic environment, we develop two incremental algorithms for this purpose (IHFSGR-1 and IHFSGR-2 for short). As a preliminary step, we propose a new uncertainty measure to quantify the amount of information contained in the hierarchical classification system, and based on this, we develop a non-incremental hierarchical feature selection algorithm. Next, we investigate the updating mechanism of this uncertainty measure upon the arrival of samples, and propose two strategies for adding and deleting features, leading to the development of two incremental algorithms. Finally, we conduct some comparative experiments with several non-incremental algorithms. The experimental results suggest that compared with several non-incremental algorithms, our incremental algorithms can achieve better performance in terms of the classification accuracy and two hierarchical evaluation metrics, and can significantly accelerate the fuzzy rough set-based hierarchical feature selection. [ABSTRACT FROM AUTHOR]

The automated classification of rock images is of paramount importance in geological analysis, as it serves as the foundational criterion for the categorization of rock lithology. Despite recent advancements in leveraging deep learning technologies to enhance the efficiency and precision of image classification, a crucial aspect has been overlooked: these methods face a performance bottleneck when attempting to apply it directly to rock classification methods. To address this limitation, we propose a multiple granularity Spatial disorder Hierarchical residual Network (SHN). This approach involves learning from objects annotated at different levels, thereby facilitating the transfer of hierarchical knowledge across levels. By enabling lower-level classes to inherit pertinent attributes from higher-level superclasses, our method aims to capture the intricate hierarchical relationships among different rock types. Especially, we introduce a multi-granularity spatial disorder module to aid neural networks in discerning discriminative details across various scales. This module enables processed images to exhibit region independence, compelling the network to adeptly identify discriminative local regions at diverse granularity levels and extract pertinent features. Furthermore, in light of the absence of a comprehensive rock dataset, this study amassed 4,227 rock images of diverse compositions from various places, culminating in the creation of a robust rock dataset for classification. Rigorous experimentation on this dataset yielded highly promising results, demonstrating the effectiveness of our proposed method in addressing the challenges of rock image classification. [ABSTRACT FROM AUTHOR]

Aims/hypothesis: There is increasing evidence for heterogeneity in type 1 diabetes mellitus (T1D): not only the age of onset and disease progression rate differ, but also the risk of complications varies markedly. Consequently, the presence of different disease endotypes has been suggested. Impaired T and B cell responses have been established in newly diagnosed diabetes patients. We hypothesized that deciphering the immune cell profile in peripheral blood of adults with longstanding T1D may help to understand disease heterogeneity. Methods: Adult patients with longstanding T1D and healthy controls (HC) were recruited, and their blood immune cell profile was determined using multicolour flow cytometry followed by a machine-learning based elastic-net (EN) classification model. Hierarchical clustering was performed to identify patientspecific immune cell profiles. Results were compared to those obtained in matched healthy control subjects. Results: Hierarchical clustering analysis of flow cytometry data revealed three immune cell composition-based distinct subgroups of individuals: HCs, T1Dgroup-A and T1D-group-B. In general, T1D patients, as compared to healthy controls, showed a more active immune profile as demonstrated by a higher percentage and absolute number of neutrophils, monocytes, total B cells and activated CD4+CD25+ T cells, while the abundance of regulatory T cells (Treg) was reduced. Patients belonging to T1D-group-A, as compared to T1D-group-B, revealed a more proinflammatory phenotype characterized by a lower percentage of FOXP3+ Treg, higher proportions of CCR4 expressing CD4 and CD8 T cell subsets, monocyte subsets, a lower Treg/conventional Tcell (Tconv) ratio, an increased proinflammatory cytokine (TNFα, IFNβ) and a decreased antiinflammatory (IL-10) producing potential. Clinically, patients in T1D-group-A had more frequent diabetes-related macrovascular complications. Conclusions: Machine-learning based classification of multiparameter flow cytometry data revealed two distinct immunological profiles in adults with longstanding type 1 diabetes; T1D-group-A and T1D-group-B. T1D-group-A is characterized by a stronger pro-inflammatory profile and is associated with a higher rate of diabetes-related (macro)vascular complications. [ABSTRACT FROM AUTHOR]

Visual localization is critical to many applications in computer vision and robotics. To address single-image RGB localization, state-of-the-art feature-based methods match local descriptors between a query image and a pre-built 3D model. Recently, deep neural networks have been exploited to regress the mapping between raw pixels and 3D coordinates in the scene, and thus the matching is implicitly performed by the forward pass through the network. However, in a large and ambiguous environment, learning such a regression task directly can be difficult for a single network. In this work, we present a new hierarchical scene coordinate network to predict pixel scene coordinates in a coarse-to-fine manner from a single RGB image. The proposed method, which is an extension of HSCNet, allows us to train compact models which scale robustly to large environments. It sets a new state-of-the-art for single-image localization on the 7-Scenes, 12-Scenes, Cambridge Landmarks datasets, and the combined indoor scenes. [ABSTRACT FROM AUTHOR]

Even if to mainstream psychologists fear is one of the seven universal emotions, discrete, measurable and with clearly distinct features, in the humanities we consider fear as a widespread concept we associate with more complex prompts than the physiological response to a hazard. This research explores various ways we describe FEAR in Spanish. For this we have made use of digital humanities tools and methods, mainly corpus linguistics and natural language processing, which enable us to explore, present and visualize linguistic elements that define fear in Mexican society. Thus we have explored this emotion (and its family: anxiety, horror, apprehension, dread, panic, terror) by examining the way it is verbalized in an ad hoc corpus covering four genres/domains: chronicle, essay, the press, and social media, specifically, tweets. A set of semantic similarity and ranking metrics were applied to the texts to identify each genre’s characteristics in association with fear. The results show that fear is an emotion that, even if it differs depending on the genre, responds to the prompts of a modern society in which danger is still being represented by illness, violence, power, or an out-group.

The dynamic changes in grassland types are crucial for conserving grassland biodiversity, conducting comprehensive health assessments, and gaining insights into ecosystem evolution. However, accurately mapping grassland types remains an ongoing challenge, especially over large areas. In response, we developed a hierarchical classification framework using random forest to tackle this task. This framework was structured into two tiers: land cover (LC) classification and grassland-type mapping, with each tier using specific features tailored to its respective objectives. In this framework, LC samples were automatically generated using existing LC products combined with vegetation indices and phenological features. We fused spectral information, phenological features, and habitat factors such as topographic indices, soil, and climate from multisource earth observation (EO) data to enhance grassland-type mapping performance. This framework successfully generated distribution maps of grassland types on the Qinghai Plateau for the years 1990–2020. Our findings revealed the following: (1) Using the two-tier classification framework and fusing multisource features, an ideal distribution map of grassland types was obtained, with a macro-average F1 score (F1) of 91% and an overall accuracy (OA) of 96%; (2) Compared to a one-shot classification framework, the two-tier classification framework achieved higher accuracy, with F1 and OA increasing by 11% and 7%, respectively; (3) Climate, topographical, soil, and phenological features assisted in distinguishing grassland types with similar spectral characteristics, especially for zonal grasslands. Adding these features increased F1 by 21%, 10%, 7%, and 3%, respectively, while OA increased by 9%, 5%, 5%, and 2%, respectively. Among these features, relative humidity, total precipitation from May to September, geographic coordinates, and elevation had the greatest effect on grassland type differentiation. June and July were the optimal phenological periods for mapping; (4) Over the 30-year period, the grasslands on the Qinghai Plateau showed an expanding trend, with extensive areas of alpine steppes transforming into alpine meadows. The results of this study provide the first elucidation of grassland-type changes across the Qinghai Plateau from 1990 to 2020. Moreover, they underscore the potential of hierarchical classification frameworks and the integration of multisource EO data in mapping grassland-type distribution.

Modern neurosurgery strives to maximize tumor removal while preserving healthy tissue integrity. Accurate intraoperative differentiation between tumor and healthy tissue is crucial yet challenging. Often neurosurgeons rely on their experience and haptic feedback during palpation to distinguish between tumor and healthy tissue. A commonly used hand-held tool for tissue removal during neurosurgery is the ultrasonic aspirator, which changes its electrical properties as it interacts with tissue. The goal is to equip the ultrasonic aspirator with the ability to differentiate between different types of tissue while at the same time not interfering with the surgical workflow and providing comprehensible outcomes. To this end, a hierarchical classification approach is employed as a proof of concept, enabling precise identification of tissue stiffness during resection.The hierarchical approach is compared with the standard flat classification, commonly used in machine learning. Within the hierarchical approach, two strategies are employed: mandatory leaf-node predictions (MLNP) and non-mandatory leaf-node predictions (NMLNP). The NMLNP allows prediction to revert to a parent node when certainty is low. Data are acquired on three artificial tissue models – differing in stiffness – with an ultrasonic aspirator in a hand-held manner. The dataset comprises 1,821 data points for training and 186 for testing after balancing.The results indicate a slight performance advantage for the hierarchical classification MLNP approach over the flat classification approach in the absence of confidence thresholds, with weighted F2-scores of 0.781 and 0.762, respectively. However, the application of confidence thresholds results in both approaches exhibiting comparable performance, with the hierarchical NMLNP approach achieving a weighted F1-score of 0.920, thereby demonstrating superior overall performance. The effects of enforcing these thresholds and excluding data with low certainty are thoroughly investigated. This work emphasizes the feasibility of tissue differentiation using a hand-held ultrasound aspirator while resecting tissue. Moreover, it highlights the capability of hierarchical classification in advancing tissue differentiation accuracy during neurosurgical procedures, which could ultimately aid surgeons and enhance the safety of intraoperative workflows.

In the article there are presented results of the study of the state of user competencies for different specialties of the university digital educational environment (UDEE) on issues related to information security (IS). The methods of cluster analysis and analysis of digital (electronic) traces (DT) of users are used. On the basis of analyzing the DTs of different groups of registered users in the UDEE, 6 types of users are identified. These types of users were a result of applying hierarchical classification and k-means method. Users were divided into appropriate clusters according to the criteria affecting IS risks. For each cluster, the UDEE IS expert can determine the probability of occurrence of high IS risk incidents and, accordingly, measures can be taken to address the causes of such incidents. The algorithms proposed in this study enable research during log file analysis aimed at identifying breaches of information security within the university's DEE.

Abstract Skin lesion classification plays a crucial role in the early detection and diagnosis of various skin conditions. Recent advances in computer-aided diagnostic techniques have been instrumental in timely intervention, thereby improving patient outcomes, particularly in rural communities lacking specialized expertise. Despite the widespread adoption of convolutional neural networks (CNNs) in skin disease detection, their effectiveness has been hindered by the limited size and data imbalance of publicly accessible skin lesion datasets. In this context, a two-step hierarchical binary classification approach is proposed utilizing hybrid machine and deep learning (DL) techniques. Experiments conducted on the International Skin Imaging Collaboration (ISIC 2017) dataset demonstrate the effectiveness of the hierarchical approach in handling large class imbalances. Specifically, employing DenseNet121 (DNET) as a feature extractor and random forest (RF) as a classifier yielded the most promising results, achieving a balanced multiclass accuracy (BMA) of 91.07% compared to the pure deep-learning model (end-to-end DNET) with a BMA of 88.66%. The RF ensemble exhibited significantly greater efficiency than other machine-learning classifiers in aiding DL to address the challenge of learning with limited data. Furthermore, the implemented predictive hybrid hierarchical model demonstrated enhanced performance while significantly reducing computational time, indicating its potential efficiency in real-world applications for the classification of skin lesions.

Skin lesion classification plays a crucial role in the early detection and diagnosis of various skin conditions. Recent advances in computer-aided diagnostic techniques have been instrumental in timely intervention, thereby improving patient outcomes, particularly in rural communities lacking specialized expertise. Despite the widespread adoption of convolutional neural networks (CNNs) in skin disease detection, their effectiveness has been hindered by the limited size and data imbalance of publicly accessible skin lesion datasets. In this context, a two-step hierarchical binary classification approach is proposed utilizing hybrid machine and deep learning (DL) techniques. Experiments conducted on the International Skin Imaging Collaboration (ISIC 2017) dataset demonstrate the effectiveness of the hierarchical approach in handling large class imbalances. Specifically, employing DenseNet121 (DNET) as a feature extractor and random forest (RF) as a classifier yielded the most promising results, achieving a balanced multiclass accuracy (BMA) of 91.07% compared to the pure deep-learning model (end-to-end DNET) with a BMA of 88.66%. The RF ensemble exhibited significantly greater efficiency than other machine-learning classifiers in aiding DL to address the challenge of learning with limited data. Furthermore, the implemented predictive hybrid hierarchical model demonstrated enhanced performance while significantly reducing computational time, indicating its potential efficiency in real-world applications for the classification of skin lesions. [ABSTRACT FROM AUTHOR]

Molecular subtyping of gastric cancer (GC) aims to comprehend its genetic landscape. However, the efficacy of current subtyping methods is hampered by their mixed use of molecular features, a lack of strategy optimization, and the limited availability of public GC datasets. There is a pressing need for a precise and easily adoptable subtyping approach for early DNA‐based screening and treatment. Based on TCGA subtypes, we developed a novel DNA‐based hierarchical classifier for gastric cancer molecular subtyping (HCG), which employs gene mutations, copy number aberrations, and methylation patterns as predictors. By incorporating the closely related esophageal adenocarcinomas dataset, we expanded the TCGA GC dataset for the training and testing of HCG (n = 453). The optimization of HCG was achieved through three hierarchical strategies using Lasso‐Logistic regression, evaluated by their overall the area under receiver operating characteristic curve (auROC), accuracy, F1 score, the area under precision‐recall curve (auPRC) and their capability for clinical stratification using multivariate survival analysis. Subtype‐specific DNA alteration biomarkers were discerned through difference tests based on HCG defined subtypes. Our HCG classifier demonstrated superior performance in terms of overall auROC (0.95), accuracy (0.88), F1 score (0.87) and auPRC (0.86), significantly improving the clinical stratification of patients (overall p‐value = 0.032). Difference tests identified 25 subtype‐specific DNA alterations, including a high mutation rate in the SYNE1, ITGB4, and COL22A1 genes for the MSI subtype, and hypermethylation of ALS2CL, KIAA0406, and RPRD1B genes for the EBV subtype. HCG is an accurate and robust classifier for DNA‐based GC molecular subtyping with highly predictive clinical stratification performance. The training and test datasets, along with the analysis programs of HCG, are accessible on the GitHub website (github.com/LabxSCUT). [ABSTRACT FROM AUTHOR]

Copyright of Dicle University Journal of Engineering / Dicle Üniversitesi Mühendislik Dergisi is the property of Dicle Universitesi and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of Chongqing University of Posts & Telecommunications (Natural Science Edition) is the property of Chongqing University of Posts & Telecommunications and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Classification and recognition tasks confronted by intelligent systems are becoming complicated as the sizes of samples, dimensionality and labels dramatically increase in the past few years. Learning machines have to deal with modeling tasks with millions of samples and ten of thousand of labels in high-dimensional feature spaces. These tasks pose great challenges for traditional learning paradigms. Inspired by the progress in cognitive and neural science, we propose an end-to-end deep hierarchical classification framework and integrate the processes of representation learning, hierarchical structure construction, and hierarchical classification modeling in this work. This integrative framework brings several advantages: (1) the learned features are good for hierarchical structure construction and classification; (2) the learned hierarchical structure is beneficial for the subsequent hierarchical classification; (3) the learned hierarchical classification model supervises and guides the representation learning and hierarchical structure construction. This new hierarchical paradigm can not only well deal with large-scale classification tasks but also provide new inspirations to other research fields of artificial intelligence. [ABSTRACT FROM AUTHOR]

In the article there are presented results of the study of the state of user competencies for different specialties of the university digital educational environment (UDEE) on issues related to information security (IS). The methods of cluster analysis and analysis of digital (electronic) traces (DT) of users are used. On the basis of analyzing the DTs of different groups of registered users in the UDEE, 6 types of users are identified. These types of users were a result of applying hierarchical classification and k-means method. Users were divided into appropriate clusters according to the criteria affecting IS risks. For each cluster, the UDEE IS expert can determine the probability of occurrence of high IS risk incidents and, accordingly, measures can be taken to address the causes of such incidents. The algorithms proposed in this study enable research during log file analysis aimed at identifying breaches of information security within the university's DEE. [ABSTRACT FROM AUTHOR]

Geodiversity is a topical concept in earth and environmental sciences. Geodiversity information is needed to conserve nature, use ecosystem services and achieve sustainable development goals. Despite the increasing demand for geodiversity data, there exists no comprehensive system for categorizing geodiversity. Here, we present a hierarchically structured taxonomy that is potentially applicable in mapping and quantifying geodiversity across different regions, environments and scales. In this taxonomy, the main components of geodiversity are geology, geomorphology, hydrology and pedology. We propose a six-level hierarchical system where the components of geodiversity are classified at progressively lower taxonomic levels based on their genesis, physical–chemical properties and morphology. This comprehensive taxonomy can be used to compile geodiversity information for scientific research and various applications of value to society and nature conservation. Ultimately, this hierarchical system is the first step towards developing a global geodiversity taxonomy. This article is part of the Theo Murphy meeting issue 'Geodiversity for science and society'. [ABSTRACT FROM AUTHOR]

Mapping land use and land cover (LULC) using remote sensing is fundamental to environmental monitoring, spatial planning and characterising drivers of change in landscapes. We develop a new, general and versatile approach for mapping LULC in landscapes with relatively gradual transition between LULC categories such as African savannas. The approach integrates a well-tested hierarchical classification system with the computationally efficient random forest (RF) classifier and produces detailed, accurate and consistent classification of structural vegetation heterogeneity and density and anthropogenic land use. We use Landsat 8 OLI imagery to illustrate this approach for the Extended Greater Masai Mara Ecosystem (EGMME) in southwestern Kenya. We stratified the landscape into eight relatively homogeneous zones, systematically inspected the imagery and randomly allocated 1,697 training sites, 556 of which were ground-truthed, proportionately to the area of each zone. We directly assessed the accuracy of the visually classified image. Accuracy was high and averaged 88.1% (80.5%-91.7%) across all the zones and 89.1% (50%-100%) across all the classes. We applied the RF classifier to randomly selected samples from the original training dataset, separately for each zone and the EGMME. We evaluated the overall and class-specific accuracy and computational efficiency using the Out-of-Bag (OOB) error. Overall accuracy (79.3%-97.4%) varied across zones but was higher whereas the class-specific accuracy (25.4%-98.1%) was lower than that for the EGMME (80.2%). The hierarchical classifier identified 35 LULC classes which we aggregated into 18 intermediate mosaics and further into five more general categories. The open grassed shrubland (21.8%), sparse shrubbed grassland (10.4%) and small-scale cultivation (13.3%) dominated at the detailed level, grassed shrubland (31.9%) and shrubbed grassland (28.9%) at the intermediate level, and grassland (35.7%), shrubland (35.3%) and woodland (12.5%) at the general level. Our granular LULC map for the EGMME is sufficiently accurate for important practical purposes such as land use spatial planning, habitat suitability assessment and temporal change detection. The extensive ground-truthing data, sample site photos and classified maps can contribute to wider validation efforts at regional to global scales. [ABSTRACT FROM AUTHOR]

The performance of emotive text classification using affective hierarchical schemes (e.g., WordNet-Affect) is often evaluated using the same traditional measures used to evaluate the performance of when a finite set of isolated classes are used. However, applying such measures means the full characteristics and structure of the emotive hierarchical scheme are not considered. Thus, the overall performance of emotive text classification using emotion hierarchical schemes is often inaccurately reported and may lead to ineffective information retrieval and decision making. This paper provides a comparative investigation into how methods used in hierarchical classification problems in other domains, which extend traditional evaluation metrics to consider the characteristics of the hierarchical classification scheme, can be applied and subsequently improve the classification of emotive texts. This study investigates the classification performance of three widely used classifiers, Naive Bayes, J48 Decision Tree, and SVM, following the application of the aforementioned methods. The results demonstrated that all the methods improved the emotion classification. However, the most notable improvement was recorded when a depth-based method was applied to both the testing and validation data, where the precision, recall, and F1-score were significantly improved by around 70 percentage points for each classifier. [ABSTRACT FROM AUTHOR]

The precise categorization of brief texts holds significant importance in various applications within the ever-changing realm of artificial intelligence (AI) and natural language processing (NLP). Short texts are everywhere in the digital world, from social media updates to customer reviews and feedback. Nevertheless, short texts' limited length and context pose unique challenges for accurate classification. This research article delves into the influence of data sorting methods on the quality of manual labeling in hierarchical classification, with a particular focus on short texts. The study is set against the backdrop of the increasing reliance on manual labeling in AI and NLP, highlighting its significance in the accuracy of hierarchical text classification. Methodologically, the study integrates AI, notably zero-shot learning, with human annotation processes to examine the efficacy of various data-sorting strategies. The results demonstrate how different sorting approaches impact the accuracy and consistency of manual labeling, a critical aspect of creating high-quality datasets for NLP applications. The study's findings reveal a significant time efficiency improvement in terms of labeling, where ordered manual labeling required 760 min per 1000 samples, compared to 800 min for traditional manual labeling, illustrating the practical benefits of optimized data sorting strategies. Comparatively, ordered manual labeling achieved the highest mean accuracy rates across all hierarchical levels, with figures reaching up to 99% for segments, 95% for families, 92% for classes, and 90% for bricks, underscoring the efficiency of structured data sorting. It offers valuable insights and practical guidelines for improving labeling quality in hierarchical classification tasks, thereby advancing the precision of text analysis in AI-driven research. This abstract encapsulates the article's background, methods, results, and conclusions, providing a comprehensive yet succinct study overview. [ABSTRACT FROM AUTHOR]

Wetland system is one of the most important ecosystems on the earth’s surface. It is significant important to monitor wetland ecosystem using remote sensing technology. However, the complexity, fuzziness, and spatial heterogeneity of wetlands increase the difficulty of wetland classification, leading to the problem that the classification accuracy is not high enough to satisfy the needs of in-depth research. At present, the classification of wetlands is mainly based on pixel based- and image object based- methods. Addressing the problems of traditional pixel based- and image object based- methods, this study proposes to utilize fuzzy geospatial objects to express wetland objects. By synthesizing the spectral features, shape features, texture features, fuzziness and other features of wetland objects, a hierarchical classification method based on fuzzy geospatial objects is proposed. Taking Tianjin Binhai New Area as the study area, Sentinel-2 satellite remote sensing images are utilized for verification. The main contents of this study and its results are as follows: (1) Extract the fuzzy geospatial objects of wetlands and construct the classification feature sets. (2) To simplify the classification problem, a hierarchical classification framework based on optimizing multiple attributes using Random Forest is proposed. By this method, the problems of difficulty in distinguishing wetlands and low classification accuracy caused by similarity of spectral features of wetland objects in the traditional single layer classification method are solved. Three experiments are designed in the study to verify the effects of the fuzzy geospatial objects of wetlands and the hierarchical classification method on the classification accuracy of wetlands, respectively. The results show that the overall accuracy and Kappa coefficient of the proposed hierarchical wetland classification method based on fuzzy geospatial objects are 94.35% and 0.899, respectively, which are 12.35% and 0.183 higher than those of the traditional image object based- methods.

The automated classification of rock images is of paramount importance in geological analysis, as it serves as the foundational criterion for the categorization of rock lithology. Despite recent advancements in leveraging deep learning technologies to enhance the efficiency and precision of image classification, a crucial aspect has been overlooked: these methods face a performance bottleneck when attempting to apply it directly to rock classification methods. To address this limitation, we propose a multiple granularity Spatial disorder Hierarchical residual Network (SHN). This approach involves learning from objects annotated at different levels, thereby facilitating the transfer of hierarchical knowledge across levels. By enabling lower-level classes to inherit pertinent attributes from higher-level superclasses, our method aims to capture the intricate hierarchical relationships among different rock types. Especially, we introduce a multi-granularity spatial disorder module to aid neural networks in discerning discriminative details across various scales. This module enables processed images to exhibit region independence, compelling the network to adeptly identify discriminative local regions at diverse granularity levels and extract pertinent features. Furthermore, in light of the absence of a comprehensive rock dataset, this study amassed 4,227 rock images of diverse compositions from various places, culminating in the creation of a robust rock dataset for classification. Rigorous experimentation on this dataset yielded highly promising results, demonstrating the effectiveness of our proposed method in addressing the challenges of rock image classification.

Time Series Classification, is one of the very challenging areas in pattern recognition problems. As the volume of time series data increases, a multitude of TSC algorithms have been developed. Notably, only a limited number of these approaches have ventured into the realm of Deep Neural Networks to tackle this task. This paper presents a new hierarchical deep neural network architecture with a multi-section learning mechanism for the classification of multi-output time series datasets. The hierarchical structure of the proposed approach uses separate segments to detect output classes of different granular levels. The presented model does not suffer from heavy computations during training found in commonly used in other deep learning models such as convolutional neural network. Additionally, our suggested method handles data dimension reduction automatically in the hidden layers of different network sections and does not require separate pre-processing units for this purpose. To demonstrate the effectiveness of the proposed model, we have utilized it in the wearable sensor-based Human Activity Recognition problem and evaluated its performance on several benchmark datasets in TSC and HAR, such as PAMAP2 and Opportunity, with several different configurations. The empirical results on the benchmark datasets show that our proposed model outperforms several other deep learning-based solutions based on measures such as $F_{1}$ -Score, accuracy, and ROC curve.

There is a growing need for systems that can be used to effectively detect and classify intrusions in extensive network data exchanges. To this end, we propose Hi-MLIC, a hierarchical multilayer lightweight intrusion classification model that has been designed to address various intrusion types. This study highlights the challenges involved in classifying intrusions due to data imbalance across different types of intrusion data along with the complex nature of consolidating multiple benchmark datasets into cohesive datasets for real-time detection. To address these issues, we consolidated packet capture data from two widely used benchmark datasets, CIC-IDS2017 and UNSW-NB15, into two newer and more comprehensive datasets, CM-CIC-IDS2017 and CM-UNSW-NB15, respectively. This consolidation enables the identification and classification of a broader range of intrusion types. Our hierarchical approach achieves improved classification accuracy by effectively addressing the class imbalance that is inherent in non-hierarchical models. Layer-1 separates network traffic into benign and malicious categories. Layer-2 further classifies malicious traffic into four groups, while Layer-3 identifies 23 specific intrusion types. We reduced the model complexity and processing time by performing misclassification analysis and eliminating unnecessary features. Our model ultimately achieved a recall metric of up to 98.8%, thus demonstrating its effectiveness and efficiency in intrusion detection and classification. Altogether, the proposed Hi-MLIC represents a significant advancement in addressing the challenges of real-time network intrusion detection.

Human Activity Recognition (HAR) based on Inertial Measurement Unit (IMU) has become increasingly important in health and fitness applications. These systems can continuously and cost-effectively monitor human activity, regardless of the surrounding environment. However, the current dominant trend in HAR uses black box flat classification (FC) methods, which lack interpretability and do not consider the natural hierarchical relationship between activity classes. Such systems often achieve greater accuracy at the cost of increased complexity and are not suitable for critical decision-making applications. This paper proposes an Adaptive Hierarchical Decision Tree (AHDT) HAR system that recognizes human activities based on IMU measurements along with quasi-stationary inclination feature extraction. The proposed system generates a global classifier that classifies human activities according to a tree taxonomy structure. This approach maintains interpretability while considering the fundamental signal data features embedded in the natural hierarchical representation of the activity classes. In addition to the commonly used flat classification accuracy measures, we applied modified hierarchical accuracy measures to assess the exclusive characteristics of hierarchical relationships between the classes. We used seven publicly available datasets to evaluate the proposed system and compared its performance with other tree-based classifiers, including Random Forest, Gradient Boosting, XGBoost, and AdaBoost classifiers. Our results demonstrated that the AHDT system significantly improves the recognition performance of fine-grained activities and offers a balance between lower complexity and higher interpretability. Overall, the proposed AHDT system provides an interpretable and practical approach to HAR that can be valuable in critical decision-making applications.

In the era of big data, the amount of class labels is growing rapidly, which poses a great challenge to classification tasks. The hierarchical classification was thus introduced to address this issue by considering the structural information between different class labels. In this paper, we propose an incremental feature selection algorithm for handling the arrival of new samples by using the theory of fuzzy rough sets. As a preliminary step, we propose a non-incremental hierarchical feature selection algorithm, which is an improved version of the existing method. Then utilizing the sibling strategy, the incremental calculation of the dependency degree at the arrival of samples is discussed. Based on the analysis of dependency degree change, we design feature addition and deletion strategies, as well as the incremental feature selection algorithm. In the experimental section, two versions of algorithms are designed. The experimental results show that our improvement of the existing method is highly effective and can significantly accelerate the process of feature selection. In addition, version 2 of the incremental algorithm exhibits much higher efficiency than the improved non-incremental algorithm on several datasets, as well as the existing method. Compared to six hierarchical feature selection algorithms, our algorithm achieves better results on the classification accuracy and three hierarchical evaluation metrics. The effectiveness and efficiency of version 1 are also verified by the comparison of version 2 and other results. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Signal Processing is the property of Journal of Signal Processing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In the big data era, the sample size is becoming increasingly large, the data dimensionality is also becoming extremely high, moreover, there exists hierarchical structure between different class labels. This paper investigates incremental feature selection for hierarchical classification based on the dependency degree of inclusive strategy and solves the hierarchical classification problem where labels are distributed at arbitrary nodes in tree structure. Firstly, the inclusive strategy is used to reduce the negative sample space by exploiting the hierarchical label structure. Secondly, a new fuzzy rough set model is introduced based on inclusive strategy, and a dependency calculation algorithm based on the inclusive strategy and a non-incremental feature selection algorithm are also proposed. Then, the dependency degree based on the inclusive strategy is proposed by adopting the incremental mechanism. Based on these, two incremental feature selection frameworks based on two strategies are designed. Lastly, a comparative study with the method based on the sibling strategy is performed. The?feasibility?and?efficiency?of the proposed algorithms are verified by numerical experiments.

The notion of style has been the subject of much research and theoretical development in prehistoric archaeology. This vast concept touches several fields, including the morphology of artefacts, technical gestures and their function. The lithic typology of arrowhead armatures was widely developed in the twentieth century for the study of the French Mesolithic and includes some of these aspects. The functional, technical, or aesthetic nature of the morphological characteristics of these artefacts has received little attention, while the mapping of types has been overvalued This article presents a new approach combining statistical analysis with the use-wear analysis of arrowhead from several Breton sites, dated to the Late Mesolithic.

Mapping land use and land cover (LULC) using remote sensing is fundamental to environmental monitoring, spatial planning and characterising drivers of change in landscapes. We develop a new, general and versatile approach for mapping LULC in landscapes with relatively gradual transition between LULC categories such as African savannas. The approach integrates a well-tested hierarchical classification system with the computationally efficient random forest (RF) classifier and produces detailed, accurate and consistent classification of structural vegetation heterogeneity and density and anthropogenic land use. We use Landsat 8 OLI imagery to illustrate this approach for the Extended Greater Masai Mara Ecosystem (EGMME) in southwestern Kenya. We stratified the landscape into eight relatively homogeneous zones, systematically inspected the imagery and randomly allocated 1,697 training sites, 556 of which were ground-truthed, proportionately to the area of each zone. We directly assessed the accuracy of the visually classified image. Accuracy was high and averaged 88.1% (80.5%–91.7%) across all the zones and 89.1% (50%–100%) across all the classes. We applied the RF classifier to randomly selected samples from the original training dataset, separately for each zone and the EGMME. We evaluated the overall and class-specific accuracy and computational efficiency using the Out-of-Bag (OOB) error. Overall accuracy (79.3%–97.4%) varied across zones but was higher whereas the class-specific accuracy (25.4%–98.1%) was lower than that for the EGMME (80.2%). The hierarchical classifier identified 35 LULC classes which we aggregated into 18 intermediate mosaics and further into five more general categories. The open grassed shrubland (21.8%), sparse shrubbed grassland (10.4%) and small-scale cultivation (13.3%) dominated at the detailed level, grassed shrubland (31.9%) and shrubbed grassland (28.9%) at the intermediate level, and grassland (35.7%), shrubland (35.3%) and woodland (12.5%) at the general level. Our granular LULC map for the EGMME is sufficiently accurate for important practical purposes such as land use spatial planning, habitat suitability assessment and temporal change detection. The extensive ground-truthing data, sample site photos and classified maps can contribute to wider validation efforts at regional to global scales.

In hierarchical classification learning, the feature space of data has high dimensionality and is unknown with emergent features. To solve the above problems, we propose an online hierarchical feature selection algorithm based on adaptive ReliefF. Firstly, ReliefF is adaptively improved via using the density information of instances around the target sample, making it unnecessary to prespecify parameters. Secondly, the hierarchical relationship between classes is used, and a new method for calculating the feature weight of hierarchical data is defined. Then, an online correlation analysis method based on feature interaction is designed. Finally, the adaptive ReliefF algorithm is improved based on feature redundancy, and the feature weight is scaled by the correlation between features in order to achieve the dynamic updating of feature redundancy. A large number of experiments verify the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

Few-shot learning aims to recognize new visual concepts with a small number of labeled samples. The hierarchical structure based on inter-class labels performs well in many few-shot learning models. However, intra-class features are similar and difficult to distinguish, which is important for mining the correlation and independence between intra-class features in the scene of sparse data. In this paper, we propose a few-shot learning model with a self-similarity feature representation by a hierarchical relation network, which considers inter-class labels and intra-class features to guide few-shot learning. First, we introduce a self-similarity feature representation module as the intermediate feature transform in the neural network. Unlike the traditional model, it extracts specific feature information from intra-class features. Second, we leverage the inter-class label hierarchical structure as important auxiliary information to establish a hierarchical relation network metric module. The module uses coarse-grained information to guide fine-grained classification, which effectively alleviates the problem of insufficient data. Experimental results show that our model improves the classification accuracy, reaching 58.68% on the tieredImageNet dataset. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Frontiers of Computer Science & Technology is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Deep neural networks (DNNs) have drawn much attention due to their success in various vision tasks. Current DNNs are used on data with a relatively small number of classes (e.g., 1000 or less) and employ a fully connected layer for classification, which allocates one neuron for each class and thus, per-example, the classification scales as O (K) with the number of classes K. This approach is computationally intensive for many real-life applications where the number of classes is very large (e.g., tens of thousands of classes). To address this problem, our paper introduces a hierarchical approach for classification with a large number of classes that scales as O (K) and could be extended to O (log K) with a deeper hierarchy. The method, called Hierarchical PPCA, uses a self-supervised pretrained feature extractor to obtain meaningful features and trains Probabilistic PCA models on the extracted features for each class separately, making it easy to add classes without retraining the whole model. The Mahalanobis distance is used to obtain the classification result. To speed-up classification, the proposed Hierarchical PPCA framework clusters the image class models, represented as Gaussians, into a smaller number of super-classes using a modified k-means clustering algorithm. The classification speed increase is obtained by Hierarchical PPCA assigning a sample to a small number of the most likely super-classes and restricting the image classification to the image classes corresponding to these super-classes. The fact that the model is trained on each class separately makes it applicable to training on very large datasets such as the whole ImageNet with more than 10,000 classes. Experiments on three standard datasets (ImageNet-100, ImageNet-1k,and ImageNet-10k) indicate that the hierarchical classifier can achieve a superior accuracy with up to a 16-fold speed increase compared to a standard fully connected classifier. [ABSTRACT FROM AUTHOR]

Hierarchical classification (HC) assigns each object with multiple labels organized into a hierarchical structure. The existing deep learning-based HC methods usually predict an instance starting from the root node until a leaf node is reached. However, in the real world, images impaired by noise, occlusion, blur, or low resolution may not provide sufficient information for the classification at subordinate levels. To address this issue, we propose a novel Semantic Guided level-category Hybrid Prediction Network (SGHPN) that can jointly perform the level and category prediction in an end-to-end manner. SGHPN comprises two modules: a visual transformer that extracts feature vectors from the input images, and a semantic guided cross-attention module that uses categories word embeddings as queries to guide learning category-specific representations. In order to evaluate the proposed method, we construct two new datasets in which images are at a broad range of quality and thus are labeled to different levels (depths) in the hierarchy according to their individual quality. Experimental results demonstrate the effectiveness of our proposed HC method. [ABSTRACT FROM AUTHOR]

The precise categorization of brief texts holds significant importance in various applications within the ever-changing realm of artificial intelligence (AI) and natural language processing (NLP). Short texts are everywhere in the digital world, from social media updates to customer reviews and feedback. Nevertheless, short texts’ limited length and context pose unique challenges for accurate classification. This research article delves into the influence of data sorting methods on the quality of manual labeling in hierarchical classification, with a particular focus on short texts. The study is set against the backdrop of the increasing reliance on manual labeling in AI and NLP, highlighting its significance in the accuracy of hierarchical text classification. Methodologically, the study integrates AI, notably zero-shot learning, with human annotation processes to examine the efficacy of various data-sorting strategies. The results demonstrate how different sorting approaches impact the accuracy and consistency of manual labeling, a critical aspect of creating high-quality datasets for NLP applications. The study’s findings reveal a significant time efficiency improvement in terms of labeling, where ordered manual labeling required 760 min per 1000 samples, compared to 800 min for traditional manual labeling, illustrating the practical benefits of optimized data sorting strategies. Comparatively, ordered manual labeling achieved the highest mean accuracy rates across all hierarchical levels, with figures reaching up to 99% for segments, 95% for families, 92% for classes, and 90% for bricks, underscoring the efficiency of structured data sorting. It offers valuable insights and practical guidelines for improving labeling quality in hierarchical classification tasks, thereby advancing the precision of text analysis in AI-driven research. This abstract encapsulates the article’s background, methods, results, and conclusions, providing a comprehensive yet succinct study overview.

The prediction of disease can facilitate early intervention, comprehensive diagnosis and treatment, thereby benefiting healthcare and reducing medical costs. While single class and multi-class learning methods have been applied for disease prediction, they are inadequate in distinguishing between primary and secondary diagnoses, which is crucial for treatments. In this paper, label distribution is suggested to describe the diagnosis, which assigns the description degree to quantify the diagnosis. Additionally, a novel hierarchical label distribution learning (HLDL) model is proposed to make fine-grained predictions based on the hierarchical classification of diseases, taking into account the relationship among diseases. The experimental results on real-world datasets demonstrate that the HLDL model outperforms the baselines with statistical significance. [ABSTRACT FROM AUTHOR]

The Digital Economy and Society Index can be considered a complex indicator that measures the digital performance of the European Union. This indicator has undergone some methodological changes, precisely to be permanently aligned with the European Union's development strategies but especially with technological development. The aim of the paper is to highlight the impact of the standard of living and its components on digital performance at the level of the European Union. Moreover, the paper will analyse the current state of digital performance at the level of the European Union. The results can serve as a policymaking tool to facilitate the implementation of the digitalisation project at the European level. In this article, hierarchical clustering was used to define groups of countries from the perspective of digitalisation. Ward's algorithm determined four homogeneous clusters that were presented in the analysis. This approach serves as a hypothesis to test the correlation between digital performance and living standards. Thus, with the help of the linear correlation coefficient, a strong positive relationship was identified between the Human Development Index and the Digital Economy and Society Index. Moreover, at the level of the Human Development Index components, it was identified that individual well-being and education correlate linearly with the Digital Economy and Society Index; therefore, the analysis was completed with a multiple regression to verify this dependence. The last part of the case study is represented by the classic linear regression model, which confirms the hypothesis that living standards influence digital performance. [ABSTRACT FROM AUTHOR]

Landform elements with varying morphologies and spatial arrangements are recognized as feature indicator of landform classification and play a critical role in geomorphological studies. Differential geometry method has been extensively applied in prior landform element research, while its efficacy in differentiating similar morphological characteristics remains inadequate to date. To reduce reliance on geomorphometric variables and increase awareness of landform patterns, geomorphons method was generated in previous study corresponding to specific landform reclassification map based on lookup table. Besides, to address the problem of feature similarity, hierarchical classification was proposed and effectively utilized for terrain recognition through the analytical strategy of fuzzy gradient features. Thus, combining the advantages of these two aspects, a hierarchical framework was proposed in this study for landform element pattern recognition considering the morphology and hierarchy factors. First, the local triplet patterns derived from geomorphons were enhanced by setting the flatness threshold, and subsequently adopted for the primary landform element recognition. Then, as geomorphic units with the same morphology possess different spatial analytical scales, the unidentified landform elements under the principle of scale adaptation were determined by calculating the spatial correlation and entropy information. To ensure the effectiveness of this proposed method, the sampling points were randomly selected from NASADEM data and then validated against a real 3D terrain model. Quantitative results of landform element pattern recognition demonstrate that our approach can reach above 77% average accuracy. Additionally, it delineates local details more effectively than geomorphons in visual assessment, resulting in a 7% accuracy improvement in overall scale. [ABSTRACT FROM AUTHOR]