Hierarchy Of Categories Research Articles

Weakly Supervised Underwater Object Real-time Detection Based on High-resolution Attention Class Activation Mapping and Category Hierarchy

Recently, deep learning-based underwater object detection technology has achieved remarkable success. However, the accuracy and completeness of dataset instance annotation are crucial for its success. The quality of underwater images is low, severe objects clustering, and occlusion, acquiring object's annotations demands substantial time and labor costs, while mis annotation and missed annotation can also degrade model performance and limit their application in practical scenarios. To address this issue, this paper presents a novel weakly supervised underwater object real-time detection method, which is divided into two subtasks: weakly supervised object localization and real-time object detection. In the weakly supervised object localization task, we design a novel category hierarchy structure network that integrates the high-resolution attention-class activation mapping algorithm to obtain high-quality object class activation maps, weaken background interference, and obtain more complete object regions. The parameterized spatial loss module is devised to enable the model to escape from local optimal solutions, thus accurately and efficiently obtaining object pseudo-detection annotation boxes. For the real-time object detection task, the single-stage detector YOLOv7 is selected as the basic detection model, and an object perception loss function is designed based on the class activation map to jointly supervise the training process. A method for filtering noisy pseudo-supervision information is proposed to enhance the pseudo-supervision information involved in training. Ablation experiments and multi-method comparison experiments were conducted on the URPC and RUOD datasets, and the results verify the effectiveness of the proposed strategy, and our model exhibits significant advantages in detection performance and detection efficiency compared to current mainstream and advanced models.

Dynamic Short-Term Perspective Estimation Based on Formal Concept Analysis

In online shopping, user perspectives transit dynamically from abstract categories to concrete subcategories within a short period. We propose a perspective-estimation system that estimates the dynamic, short-term perspectives of users by inferring a hierarchy of categories based on their actions. The proposed system analyzes the wish list rankings of users and their operational histories to extract the categories emphasized at that moment. It then employs formal concept analysis to infer the hierarchical structure of categories, thereby visualizing the dynamic short-term perspective. In experiments involving 57 participants, the proposed method rates its match with user perspectives on a seven-point scale, achieving an average score of 4.84, outperforming the feature estimation method using latent Dirichlet allocation (LDA), which scored 4.36. The statistical significance was confirmed through the Wilcoxon rank-sum test with a statistic W=4.80 and a p-value of 1.56×10-6. Compared with LDA, the proposed system is statistically significant in terms of the degree of agreement with the perspectives.

Hierarchical cavitation intensity recognition using Sub-Master Transition Network-based acoustic signals in pipeline systems

Deep learning (DL) has emerged as the most effective approach for cavitation intensity recognition. However, current DL-based models face limitations due to the inherent assumption that all target classes are treated equally. In particular, when certain classes are more difficult to distinguish than others, or when classes can be structured into a hierarchy of categories, DL-based cavitation recognition models struggle to achieve optimal performance. In this study, we propose a hierarchical cavitation intensity recognition framework using the Sub-Master Transition Network (SMTNet) to improve the performance of cavitation intensity recognition of acoustic signal valves. The SMTNet model achieves cavitation intensity recognition from coarse to fine according to the hierarchical properties of cavitation states, overcoming the limitations of DL methods. We also present a global filter layer to address for the long-term dependence of the acoustic signals. In addition, we introduce a 1D double hierarchical residual block to capture sensitive features of acoustic signals. Moreover, a hierarchical multi-label tree is embedded in the SMTNet to map the semantic structure of cavitation states. The proposed model has been evaluated on three real-world cavitation datasets from SAMSON AG, all showing superior results and outperforming SOTA DL methods.

«У-син» уголовно-процессуального исследования преступлений в кибернетическом пространстве

The article considers the paradigm of criminal procedural investigation of crimes in cybernetic space. The composition of the named paradigm is defi ned, which includes fi ve basic elements forming a hierarchy of categories: "information", data", "knowledge", "trace" and "proof". At the same time, each of the top-level categories is based on all the previous ones and uses its properties. The key features of the selected elements due to the composition and structure of the cybernetic space are considered.

Guide for Combining Primary Tumors for Statistical Analysis in Rodent Carcinogenicity Studies.

The Tumor Combination Guide was created at the request of the U. S. Food and Drug Administration (FDA) by a Working Group of biopharmaceutical experts from international societies of toxicologic pathology, the Food and Drug Administration (FDA), and members of the Standard for Exchange of Nonclinical Data (SEND) initiative, to assist pharmacology/toxicology reviewers and biostatisticians in statistical analysis of nonclinical tumor data. The guide will also be useful to study and peer review pathologists in interpreting the tumor data. This guide provides a higher-level hierarchy of tumor types or categories correlating the tumor names from the International Harmonization of Nomenclature and Diagnostic Criteria (INHAND) publications with those available in the NEOPLASM controlled terminology (CT) code list in SEND. The version of CT used in a study should be referenced in the nonclinical study data reviewer's guide (SDRG) (section 3.1) of electronic submissions to the FDA. The tumor combination guide instructions and examples are in a tabular format to make informed decisions for combining tumor data for statistical analysis. The strategy for combining tumor types for statistical analysis is based on scientific criteria gleaned from the current scientific literature; as SEND and INHAND terminology and information evolve, this guide will be updated.

Design and Development of an Automated Product Categorization Software: AI-Driven Solutions for E-Commerce Platforms

This article outlines the design and development process of an automatic product categorization software intended for use in e-commerce and online marketplace platforms. The project aims to tackle the urgent issue of effectively classifying a wide range of products within digital markets. By utilizing artificial intelligence (AI) and machine learning methodologies, the software effectively analyzes product descriptions, enabling users to seamlessly incorporate products through automated categorization. The significance of the project is rooted in its ability to guarantee the accuracy and consistency of category hierarchies, automate the process of categorizing, and improve the overall user experience. The main goals involve the process of project planning, data collecting, and preparation. Machine learning models have been built and subsequently incorporated to facilitate the study of product descriptions. Through strict evaluation and optimization processes, a high level of accuracy and efficiency is achieved, resulting in several anticipated benefits. These benefits cover automated product categorization, enhanced user experience, and the potential for online platforms to gain a competitive edge. The key elements of innovation involve AI-driven textual analysis, learning methodologies grounded in data, and the ability to adapt to diverse industry contexts. Precautions and backup strategies are implemented to tackle technical issues, including the selection of machine learning libraries and algorithms, ensuring data quality, and integrating with various platforms. The success criteria encompass the objective of achieving a minimum prediction accuracy rate of 90%, optimizing business efficiency, enhancing user pleasure, and ensuring smooth system functioning. This project is a significant contribution to the field of product categorization inside the digital marketplace, as it provides automation, accuracy, and efficiency, ultimately resulting in an enhanced user experience.

On the conceptual nature of hybrid adverbials in Isu

We review Kießling’s (2011) assessment of the emergence of hybrid adverbials from serial verb constructions in Isu of Bantoid and the Narrow Grassfield group of East Benue Congo. He posits two paths, preverbal and postverbal, that coverb elements follow as they abandon verbal properties pertaining to syntax, morphology, and tone. Our focus rests on whether this impressive array of grammatical change might reflect a particular system of semantic concepts. We contend that one lens through which to view such a system consists of a merger of Dixon’s secondary concept types and Nuyts’ hierarchy of categories conveying clausal modification. To support our contention, we apply the resulting system of semantic concepts to Isu preverbal hybrid adverbials, finding eight conceptual types oriented toward characterization of grammatical subject. Despite the highly tentative nature of this effort, results align with a similar range of conceptual types that occur as preverbs in West Benue Congo Yoruboid and Edoid, which have also grammaticalized from verbs.

Towards human-compatible XAI: Explaining data differentials with concept induction over background knowledge

Concept induction, which is based on formal logical reasoning over description logics, has been used in ontology engineering in order to create ontology (TBox) axioms from the base data (ABox) graph. In this paper, we show that it can also be used to explain data differentials, for example in the context of Explainable AI (XAI), and we show that it can in fact be done in a way that is meaningful to a human observer. Our approach utilizes a large class hierarchy, curated from the Wikipedia category hierarchy, as background knowledge. To make the explanations easily understandable for non-specialists, the complex description logic explanations generated by our concept induction system (ECII) were presented as a word list consisting of the concept names occurring in the highest rated system responses.

CEHMR: Curriculum learning enhanced hierarchical multi-label classification for medication recommendation

The medication recommendation (MR) or medication combination prediction task aims to predict effective prescriptions given accurate patient representations derived from electronic health records (EHRs), which contributes to improving the quality of clinical decision-making, especially for patients with multi-morbidity. Although in recent years deep learning technology has achieved great success in MR, the performance of current multi-label based MR solutions is unsatisfactory. They mainly focus on improving the patient representation module and modeling the medication label dependencies such as drug–drug interaction (DDI) correlation and co-occurrence relationship. However, the hierarchical dependency among medication labels and diversity of difficulty among MR training examples lack sufficient consideration. In this paper, we propose a framework of Curriculum learning Enhanced Hierarchical multi-label classification for MR (CEHMR). Motivated by the category hierarchy of medications which organizes standard medication codes in a hierarchical structure, we utilize it to provide more trustworthy prior knowledge for modeling label dependency. Specifically, we design a hierarchical multi-label classifier with a learnable gate fusion layer, to simultaneously capture the level-independent (local) and level-dependent (global) hierarchical information in the medication hierarchy. In addition, to overcome the diversity of training example difficulties, and progressively achieve a smoother training process, we introduce a bootstrap-based curriculum learning strategy. Hence, the example difficulty can be measured based on the predictive performance of the MR model, and then all training examples would be retrained from easy to hard under the guidance of a predefined training scheduler. Experiments on the real-world medical MIMIC-III database demonstrate that the proposed framework can achieve state-of-the-art performance compared with seven representative baselines, and extensive ablation studies validate the effectiveness of each component of CEHMR.

CAT: a coarse-to-fine attention tree for semantic change detection

Semantic change detection (SCD) and land cover mapping (LCM) are always treated as a dual task in the field of remote sensing. However, due to diverse real-world scenarios, many SCD categories are not easy to be clearly recognized, such as “water-vegetation” and “water-tree”, which can be regarded as fine-grained differences. In addition, even a single LCM category is usually difficult to define. For instance, some “vegetation” categories with litter vegetation coverage are easily confused with the general “ground” category. SCD/LCM becomes challenging under both challenges of its fine-grained nature and label ambiguity. In this paper, we tackle the SCD and LCM tasks simultaneously by proposing a coarse-to-fine attention tree (CAT) model. Specifically, it consists of an encoder, a decoder and a coarse-to-fine attention tree module. The encoder-decoder structure extracts the high-level features from input multi-temporal images first and then reconstructs them to return SCD and LCM predictions. Our coarse-to-fine attention tree, on the one hand, utilizes the tree structure to better model a hierarchy of categories by predicting the coarse-grained labels first and then predicting the fine-grained labels later. On the other hand, it applies the attention mechanism to capture discriminative pixel regions. Furthermore, to address label ambiguity in SCD/LCM, we also equip a label distribution learning loss upon our model. Experiments on the large-scale SECOND dataset justify that the proposed CAT model outperforms state-of-the-art models. Moreover, various ablation studies have demonstrated the effectiveness of tailored designs in the CAT model for solving semantic change detection problems.

The Musical Turn in Biosemiotics

Human music and language are two systems of communication and expression that, while historically considered to overlap, have become increasingly divergent in their approach and study. Music and language almost certainly co-evolved and emerged from the same semiotic field, and this relationship as well as co-origin are actively researched and debated. For the sake of evaluating the semiotic content of zoomusicology, we investigate music from a ‘bottom-up’ biosemiotic functionalist account considering iconic, indexical, and symbolic forms of meaning not in a hierarchy but according to their effects on agents. Such an approach avoids overintellectualizing the representational aspects of music, and instead inverts, as it were, the traditional hierarchy of semiotic categories to produce illocutionary effects. Understanding aesthetics and action not as a priori separate but rather fundamentally co-arising elements of the same events. The focus of musicality again returns to interpretation and how semiosis precipitates expression.

TME: Tree-guided Multi-task Embedding Learning towards Semantic Venue Annotation

The prevalence of location-based services has generated a deluge of check-ins, enabling the task of human mobility understanding. Among the various types of information associated with the check-in venues, categories (e.g., Bar and Museum ) are vital to the task, as they often serve as excellent semantic characterization of the venues. Despite its significance and importance, a large portion of venues in the check-in services do not have even a single category label, such as up to 30% of venues in the Foursquare system lacking category labels. We, therefore, address the problem of semantic venue annotation, i.e., labeling the venue with a semantic category. Existing methods either fail to fully exploit the contextual information in the check-in sequences, or do not consider the semantic correlations across related categories. As such, we devise a Tree-guided Multi-task Embedding model (TME for short) to learn effective representations of venues and categories for the semantic annotation. TME jointly learns a common feature space by modeling multi-contexts of check-ins and utilizes the predefined category hierarchy to regularize the relatedness among categories. We evaluate TME over the task of semantic venue annotation on two check-in datasets. Experimental results show the superiority of TME over several state-of-the-art baselines.

Epistemic Explanations: A Theory of Telic Normativity, and What it Explains

<i>Epistemic Explanations: A Theory of Telic Normativity, and What it Explains</i>

“There’s Something There in That Hyphen”: The Lived Experiences of Asian and Asian American Higher Education Students in the Southwest Borderlands of the United States

For centuries, Asians living in the U.S. have had to negotiate between the narratives that dominant society has imposed upon them and their understanding of what it means to be Asian and Asian American. When combined with the hierarchies of racial categories, the narratives underlying monoracialism are inherently limiting, obscuring their nuanced experiences, and stripping them of their ability to express the personal constructions of their identity The purpose of this qualitative case study was to elevate the voices of Asians and Asian Americans, their process of “inventing” their identity, and how their conceptualizations begin to deconstruct and challenge monoracialism. I argue that Asians and Asian Americans engage in a process where the interpretation and revision of meaning that emerges during interactions with others can illuminate the role of master narratives and how they negotiate between these structural factors and their ideas of what it means to be Asian or Asian American. The findings suggest a negotiation between master narratives at the macro-, meso-, and micro-societal levels that help them understand what it means to be Asian and Asian American.

Hierarchical neural network with efficient selection inference

The image classification precision is vastly enhanced with the growing complexity of convolutional neural network (CNN) structures. However, the uneven visual separability between categories leads to various difficulties in classification. The hierarchical structure of categories can be leveraged to deal with it, but a few CNNs pay attention to the character of data. Besides, a network model with a hierarchical structure is promising to extract more specific features from the data than current CNNs, since, for the latter, all categories have the same fixed number of layers for feed-forward computation. In this paper, we propose to use category hierarchies to integrate ResNet-style modules to form a hierarchical network model in a top-down manner. To extract abundant discriminative features and improve the computation efficiency, we adopt residual block selection based on coarse categories to allocate different computation paths. Each residual block works as a switch to determine the JUMP or JOIN mode for an individual coarse category. Interestingly, since some categories need less feed-forward computation than others by jumping layers, the average inference time cost is reduced. Extensive experiments show that our hierarchical network achieves higher prediction accuracy with similar FLOPs on CIFAR-10 and CIFAR-100, SVHM, and Tiny-ImageNet datasets compared to original residual networks and other existing selection inference methods.

A generalized classification and coding system of Human Disease Animal Model Resource data with a case study to show improving database retrieval efficiency.

Currently there is no unified data classification and coding standard for the existing human disease animal model resource data worldwide. Different data classification and coding systems produce different retrieval methods. Some of these methods are inefficient and difficult to use. This research investigated the rules for the classification and coding of such data based on the Replication Methodology of Animal Models for Human Disease, the Classification and Coding Rules for Health Information Data Set (WS/T 306-2009), the Science and Technology Resource Identification (GB/T 32843-2016), the Scientific Data Management Measures (000014349/2018-00052), and The Generic Description Specification for Natural Science and Technology Resources. This research aimed to develop a classification and coding system for data obtained from human disease animal model resource based on the Internet environment to provide a standardized and unified foundation for the collection, saving, retrieval, and sharing of data from this resource. A complete data classification and coding table compiled in the form of letters and numbers was produced, with a classification infrastructure that expanded layer by layer according to the three dimensions (namely, system diseases, animal species, and modeling methods) and essential attributes. When necessary, it adopted the hierarchy of major, intermediate, and minor categories for certain layer and also one-to-one matched the code and classification result. Through this study, a sharing and joint construction mechanism for data from this resource can be developed by all research institutes in this field. As a case study, this research also offered technical support for constructing the database for the National Human Disease Animal Model Resource Center. The technological innovation of this paper is to derive a research oriented retrieval method, which provides technical support for the research on the current COVID-19 epidemic and on possible future epidemics.

Существовала ли вавилонская философия?

The article criticizes the theory of the assyriologist M. van de Mieroop about the existence of Babylonian philosophy, set forth in his monograph and detailed theses. Each thesis of this theory is quoted and analyzed in detail. The differ­ences between the philosophical statement and the philosophical worldview from the religious and scientific statements of the peoples of ancient Mesopotamia are determined. It has been established that a philosophical statement can only be the author’s and is necessarily criticized by other authors. In philosophical rea­soning about the universe, the principle of non-possessiveness of the universal is obligatory. The universe is understood by philosophers as the unity of all things. At the same time, the universe is conceived as consisting of primary elements. In philosophical thinking there is a hierarchy of categories, the idea of the begin­ning, the limit and the infinite. Unlike the philosophers, the authors of the Sume­rian and Akkadian texts were nameless and did not quote or criticize each other. In their reasoning, the universal is always associated with the belonging of many things to one subject, and the world order is understood exclusively as power. The universe appears in the Sumero-Akkadian texts as a multitude of possessed things. Cuneiform texts up to the 5th century B.C. do not know the concept of primary elements. They lack logical definitions and chains of logical reason­ing, the hierarchy of categories, the idea of the beginning, the limit and the infi­nite. Thus, the constructions of the priests and scientists of ancient Mesopotamia, from our point of view, cannot be called philosophical texts.

A Dynamic Web-based Application for the Classification of Vascular Anomalies

The International Society for the Study of Vascular Anomalies (ISSVA) released the most recent version of the classification of vascular anomalies in 2018.1 This revision of the 2014 classification expanded the scope of clinical entities and included molecular diagnostics. The 2018 report incorporated recent discoveries in the field and was released as a 20-page PDF file navigable through embedded links. The links lead users through the classification hierarchy, and each page lists anomalies, their corresponding genetic mutations (if known), and other details. Further information is found in five appendices in the last seven pages of the report. The document is comprehensive, but the structure has drawbacks in usability, organization, and sustainability. The format is rigid and difficult to update. The vascular anomaly field is new and ever growing. In just the past few years, there have been significant advances in discovering new clinical entities and uncovering the genetic basis for these conditions. The PDF-based classification cannot be updated routinely without reformatting entirely. Therefore, this format is unsustainable as new discoveries are made. The PDF format classification also limits organization, ease of use, and navigability. It includes a hierarchy of anomalies and categories organized into tables with internal page links that are difficult to navigate. The embedded links guide users to various pages throughout the document, which provides a nonlinear and somewhat confusing user experience. Additionally, the format of each page was unique (to accommodate custom information), and this lack of standardization stifles the user experience because of the inconsistent location of relevant information. Finally, the indices store information that cannot be placed on the appropriate PDF pages (because of space limitations), which makes navigating back and forth tedious. The aim for a new format was to streamline the design and solve the limitations of the current PDF format. We engineered an interactive web app format, which sustainably supports future updates. The classification is organized into a tree, with nested expandable nodes containing the anomalies in a hierarchical structure, similar to that of a file explorer. With this format, a user can expand the tree to view individual anomalies, associated information, and the entire hierarchy in a common location. The visual hierarchy provides a natural organization of the classification for a more intuitive experience. The app has a consistent format throughout, and it is easily modified. Minor additions and changes can be implemented quickly, without altering or compromising the framework of the format. The web app supports including additional anomaly-specific data through the use of popover buttons, which improves organization while providing comprehensive information. Popovers may contain clinical and radiological images (Figure 1), footnotes for unusual or exceptional cases (Figure 2), genomic information, and others if necessary. Additionally, we included search functionality via autocomplete “smart” select boxes (Figure 3). The user can search for a specific term by selecting it from an alphabetical list, or by typing into a free text field that limits selection to specific options dynamically. When searching for a specific anomaly, the web app opens the tree to the selected entry while briefly highlighting the entry itself and its parent nodes. Separate selectable dialog boxes allow the user to search genes and abbreviations without navigating away (Figure 4).Figure 1.: Expanded tree view of vascular anomalies classification hierarchy. Image popover shows clinical image of infantile hemangioma.Figure 2.: Popover with additional information about rapidly involuting congenital hemangioma.Figure 3.: Search function opens the tree to the selectable term (arteriovenous malformation in this case), automatically scrolls down the page, and highlights hierarchy.Figure 4.: Gene Search and Abbreviation Search dialog boxes with option filtering and autofill. The search buttons remain affixed to the bottom for easy look-up even as one scrolls up and down the tree. One can lookup terms without losing one’s place in the tree.The tree is rendered dynamically from a single data structure, which allows for flexible modification and efficiency. When data are changed it is instantly reflected in the classification. This makes minor or major alterations feasible and practical, in contrast to labor-intense reformats of PDF releases. The vascular anomaly field is changing rapidly with the recognition of new clinical entities and the identification of genes responsible for disease. This app format allows for real-time classification updates. The web app was created using React, a JavaScript library. It is a mobile-friendly application currently hosted at https://vascularanomalies.net. A pure client-side program, it can be embedded easily on the ISSVA website and updated to reflect any future changes rapidly.

“I am strong and I can get on with my life”: The subjective experience of recovery of patients treated for depression

ABSTRACT Depression has a high prevalence throughout the world, and its management and recovery still constitute a challenge for mental health professionals. Objective : The aim of the study was to characterize the subjective experience of recovery from depression based on the perspective of those who suffer from it. Method : Forty participants from two South American countries, who had been or were currently being treated for depression, took part in semi-structured and in-depth interviews. Most participants were female (78%), with ages ranging from 22 to 63 years. Interviews were analyzed using Grounded Theory, creating a hierarchy of categories that represent participants’ experience of recovery. The categories were subsequently organized around an emergent central phenomenon. Results : “Transformation of the experience of the depressed self” was constructed as the main phenomenon that accounts for the subjective understanding of recovery. This transformation consists in an increase in self-acceptance, self-appreciation, and auto-biographical contextualization, coupled with an increase in agency and empowerment. Conclusion : Recovery is experienced as a multidimensional process that goes beyond the absence of symptoms. Change is experienced as a result of active self-management and commitment. The relevance of person-centered perspectives and their subjectivity for managing depression is discussed.

High quality care following orthopaedic injury in Zambia: A qualitative, patient-centred study

BackgroundInjuries are a significant cause of mortality and morbidity, particular in low- and middle-income countries (LMICs). While there is a focus on increasing injury care capacity, less attention is given to assessing, improving, and understanding the quality of care provided, especially from a patient perspective. This study therefore aims to understand what patients from a Zambian orthopaedic ward believe good quality care to be, to identify its key components, and contribute to better understanding what patients believe local healthcare priorities could be. MethodsPatients admitted to the orthopaedic ward of a Zambian tertiary care hospital were invited to take part in-depth face-to-face interviews. Interviews were continued until thematic saturation was achieved. Interviews were recorded and transcribed. Analysis was done using an inductive grounded theory approach. ResultsOf 13 patients approached, 12 consented to take part. Analysis of the themes from the transcripts led to the emergence of four core categories of quality care which are important to the patient: i) restoring the patient to normality (category: ‘restoring normality’), ii) establishing trust between patients and providers (‘trusting the provider’), iii) respecting the patient and allowing them to maintain autonomy (‘autonomy and respect’) iv) finding ways for patients to enjoy their time in the hospital (‘enjoying life’). From these results, a patient perspective theory of quality care emerged. This theory posits the idea that high-quality care in this context needs to fulfil these four core categories. Additionally, these core categories were ranked on significance and priority. ConclusionThe hierarchy of core categories could help to identify areas to improve care quality in this setting. Not only has this study helped to determine local priorities for achieving high-quality care but can encourage others to test injured patient perceptions of care quality in comparable settings.