Supervision Methods Research Articles

Hierarchy-Aware and Label Balanced Model for Hierarchical Text Classification

Hierarchical text classification, where labels can be modeled as a hierarchical structure, is a special multi-label text classification sub-task. Current methods mainly improve model performance by modeling label dependencies. The model HGCLR with excellent performance obtains hierarchy-aware text representations through contrastive learning, but the degree of hierarchical awareness is insufficient. To address this problem, a new multi-label negative supervision method is proposed to drive text representations of samples with more different labels farther away. In addition, to tackle the problem of label imbalance in hierarchical text classification, asymmetric loss is employed to compute the classification loss, so that the model focuses on learning from difficult samples and the contributions of positive and negative labels to the loss function tend to be balanced. In summary, based on HGCLR, by adding multi-label negative supervision and replacing the classification loss with asymmetric loss, we propose a hierarchy-aware and label balanced model (HALB) for hierarchical text classification. Experimental results demonstrate that HALB outperforms several classical models for hierarchical text classification. HALB achieves the best results on both Micro-F1 and Macro-F1 and obtains average improvements of 0.51% and 1.28% respectively on four datasets compared with the strongest baseline model HGCLR.

Educating custom officers supervisors: how do they experience their role?

PurposeThis study aims to examine customs officers’ perception of the role as supervisor before and after completing formal training in practice supervision. Owing to changes in the education of customs officers, their role as supervisors has changed, necessitating formal training.Design/methodology/approachA questionnaire with three open-ended questions was given to all participants to answer before and immediately after the training. Responses were given anonymously. Additionally, free text of the participants’ reflections on the supervision method and challenges within supervision was analysed.FindingsFour categories, including subcategories, emerged from the thematic analysis: competence in practice supervision; socialising and integration; motivation; and frame conditions. Analysis of the feedback shows that the education is experienced/perceived as a contribution to their personal development, learning and motivation, leading to more reflection on their own practice and a stronger focus on contributing to the students’ learning process and shaping them as future colleagues.Originality/valueTo the authors’ knowledge, there are no similar studies within this group, and the results of this study may lead to greater awareness regarding the supervisor role and the quality of customs and border management education and, thereby, higher quality in professional practice as customs officers.

Analisis Supervisi Pengajaran

Teaching supervision is a critical aspect of effective school leadership. A school principal who is competent in teaching supervision has the ability to understand individual teacher needs and challenges, and is able to provide the necessary support and guidance. Teaching supervision is not only limited to assessing performance, but also involves efforts to improve teaching skills and student learning outcomes. Good supervision methods involve regular classroom observations, providing constructive feedback, and providing additional training or resources when needed. In addition, a school principal who is effective in teaching supervision must be able to create an environment where teachers feel supported and appreciated, so that teachers are motivated to continue to improve the quality of teaching.

Enhancing supervision efficiency and satisfaction: Evaluating the impact of the e-monprakerin model on vocational high school apprenticeship monitoring

This study presents a novel approach to addressing the challenges brought about by Industry 4.0 in vocational education, particularly within the context of Indonesia's evolving industrial landscape. With industries increasingly embracing digitalization and automation, traditional apprenticeship supervision methods have struggled to keep pace. In response, this research introduces the e-monPrakerin model, an electronic-based supervision system tailored for vocational high schools (Bahasa: Sekolah Menengah Kejuruan/SMK). Through rigorous evaluation employing a randomized control group pretest-posttest design, the study demonstrates the effectiveness of the e-monPrakerin model compared to conventional methods. Results reveal significant improvements in practicality and user satisfaction, with practicability scores soaring from 41.1% to 81.3% and user satisfaction scores escalating from 31.6% to 84.4% post-implementation. These outcomes underscore the e-monPrakerin model's ability to optimize apprenticeship monitoring, offering a cost-effective, user-friendly, and efficient solution for navigating the complexities of Industry 4.0. By leveraging information technology to bridge the gap between vocational education and industry demands, this research sheds light on the practical implications of digital monitoring solutions. Ultimately, the e-monPrakerin model represents a substantial advancement in vocational education, showcasing the efficacy of technology-driven approaches in enhancing monitoring efficiency and user satisfaction. This contribution adds valuable insights to the field, benefiting educators, policymakers, and stakeholders alike in their efforts to equip the workforce with the skills necessary to thrive in the era of Industry 4.0.

О методическом обеспечении совершенствования государственного надзора за состоянием промышленной безопасности на объектах ведения горных работ, металлургических производств и обращения взрывчатых материалов

Since the 2000s, the activities of the state supervision over industrial safety at hazardous production facilities of mining and metallurgical production in Russia have been reoriented. Fiscal methods in control and supervision practices are substituted with preventive measures, which outlines the development of new approaches to the supervision organization. The modern methodology aiming to improve the state mining supervision has been historically based on the Russian system of scientific and technical methods and organizational and administrative measures aiming at sustainable and long-term improvement of industrial safety level at hazardous production facilities of the Russian metallurgical and mining industry in conditions of internal production growth and external sanctions. The long-term experience of ensuring industrial safety at hazardous production facilities of the mining and metallurgical industry provides the key methodological ways to improve the state mining supervision: i.e., verification of reliability of identification of hazardous production facilities; broad public awareness of causes and consequences of accidents and occupational injuries and the necessary preventive measures; identification and elimination of intrasystem shortcomings in the control and supervision activities; updating and improving industrial safety norms and rules. The efficiency of such methods is tested with practical activities and the successful reduction of accident and injury rates at the facilities under supervision. The introduced mandatory requirements as defined based on the results of control and supervision activities and the analysis of accidents and fatal injuries as well as the strict compliance with such requirements have gradually led to the reduction of the accident and injuries level. This can be considered the main achievement of the methodological activity of the organization of the state supervision over the industrial safety condition at the facilities under the supervision. This also indicates the quality of the developed methods of the state supervision and their applicability for the further control, preventive, licensing and permitting, and normative activities. In recent years, new planning and checking methods based on indicators of risk of mandatory industrial safety requirements violation have been developed and implemented.

Evolutionary Tripartite Game Model in the Coal Mine Risk-Hazard Double Pre-Control and Management in China

The classification, investigation, and management of the safety risk (namely, risk-hazard double pre-control management mechanism) in coal mines were introduced into a behavioral evolution game to assess the decision-making processes of coal mine workers, regulators, and supervisors. A tripartite game tree and a revenue perception matrix were constructed, and the evolutionary stability strategy (ESS) achieved by the evolutionary system under different conditions was obtained according to the replication dynamic equation (RD). The numerical simulation of the evolution game was conducted in MATLAB, which verified the influence of various parameters on the behaviors adopted by all parties. The results show: (1) When the safety, regulatory, and supervisory costs are high, the average expected value of mine workers, regulators, and supervisors to choose safety behaviors is low, so they adopt an unsafe behavior. By contrast, reducing costs can encourage the adoption of safety behaviors. (2) When the punishment for accident losses caused by unsafe behavior and the lack of regulations and supervision is higher than 50% of the initial state, the actors in the tripartite game are more inclined to avoid risks when facing cost losses in the short term, which can lead them to choose a safety behavior. (3) When the actors in the tripartite game face different accident probabilities and accident loss, with the severity of the accident increasing to 80% of the initial state, mine workers and regulators gradually adopt safety and regulation behaviors, and supervisors quickly adopt supervision methods.

Mixed Learning- and Model-Based Mass Estimation of Heavy Vehicles

This research utilized long short-term memory (LSTM) to oversee an RLS-based mass estimator based on longitudinal vehicle dynamics for heavy-duty vehicles (HDVs) instead of using the predefined rules. A multilayer LSTM network that analyzed parameters such as vehicle speed, longitudinal acceleration, engine torque, engine speed, and estimated mass from the RLS mass estimator was employed as the supervision method. The supervisory LSTM network was trained offline to recognize when the vehicle was operated so that the RLS estimator gave an estimate with the desired accuracy and the network was used as a reliability flag. High-fidelity simulation software was employed to collect data used to train and test the network. A threshold on the error percentage of the RLS mass estimator was used by the network to check the reliability of the algorithm. The preliminary findings indicate that the reliability of the RLS mass estimator could be predicted by using the LSTM network.

Improving Semi-Supervised Text Classification with Dual Meta-Learning

The goal of semi-supervised text classification (SSTC) is to train a model by exploring both a small number of labeled data and a large number of unlabeled data, such that the learned semi-supervised classifier performs better than the supervised classifier trained on solely the labeled samples. Pseudo-labeling is one of the most widely used SSTC techniques, which trains a teacher classifier with a small number of labeled examples to predict pseudo labels for the unlabeled data. The generated pseudo-labeled examples are then utilized to train a student classifier, such that the learned student classifier can outperform the teacher classifier. Nevertheless, the predicted pseudo labels may be inaccurate, making the performance of the student classifier degraded. The student classifier may perform even worse than the teacher classifier. To alleviate this issue, in this paper, we introduce a dual meta-learning ( DML ) technique for semi-supervised text classification, which improves the teacher and student classifiers simultaneously in an iterative manner. Specifically, we propose a meta-noise correction method to improve the student classifier by proposing a Noise Transition Matrix (NTM) with meta-learning to rectify the noisy pseudo labels. In addition, we devise a meta pseudo supervision method to improve the teacher classifier. Concretely, we exploit the feedback performance from the student classifier to further guide the teacher classifier to produce more accurate pseudo labels for the unlabeled data. In this way, both teacher and student classifiers can co-evolve in the iterative training process. Extensive experiments on four benchmark datasets highlight the effectiveness of our DML method against existing state-of-the-art methods for semi-supervised text classification. We release our code and data of this paper publicly at https://github.com/GRIT621/DML.

The Power of Public Oversight: The Effects of Environmental Agency Information Disclosure on Environmental Investment by Polluting Enterprises

Abstract Collusion between government and business presents a significant challenge to environmental governance, yet academic literature rarely addresses effective measures to address this key problem. This study addresses this gap by leveraging government information disclosure data from urban environmental protection agencies in China to empirically examine the impact of such disclosures on environmental investments by Chinese polluting enterprises. Our analysis reveals the positive effects of government information disclosure by enforcement agencies on environmental investments, particularly for enterprises characterized by higher degrees of government-business collusion. Unlike traditional supervision methods from higher authorities to lower ones, government information disclosure operates under public oversight, which can mitigate shortcomings in supervision scope and agency costs. Thus, the findings of this study offer valuable insights for addressing pollution governance challenges. Keywords: Government information disclosure, Collusion, Environmental investment, Polluting enterprises.

CardSegNet: An adaptive hybrid CNN-vision transformer model for heart region segmentation in cardiac MRI

Cardiovascular MRI (CMRI) is a non-invasive imaging technique adopted for assessing the blood circulatory system’s structure and function. Precise image segmentation is required to measure cardiac parameters and diagnose abnormalities through CMRI data. Because of anatomical heterogeneity and image variations, cardiac image segmentation is a challenging task. Quantification of cardiac parameters requires high-performance segmentation of the left ventricle (LV), right ventricle (RV), and left ventricle myocardium from the background. The first proposed solution here is to manually segment the regions, which is a time-consuming and error-prone procedure. In this context, many semi- or fully automatic solutions have been proposed recently, among which deep learning-based methods have revealed high performance in segmenting regions in CMRI data. In this study, a self-adaptive multi attention (SMA) module is introduced to adaptively leverage multiple attention mechanisms for better segmentation. The convolutional-based position and channel attention mechanisms with a patch tokenization-based vision transformer (ViT)-based attention mechanism in a hybrid and end-to-end manner are integrated into the SMA. The CNN- and ViT-based attentions mine the short- and long-range dependencies for more precise segmentation. The SMA module is applied in an encoder-decoder structure with a ResNet50 backbone named CardSegNet. Furthermore, a deep supervision method with multi-loss functions is introduced to the CardSegNet optimizer to reduce overfitting and enhance the model’s performance. The proposed model is validated on the ACDC2017 (n=100), M&Ms (n=321), and a local dataset (n=22) using the 10-fold cross-validation method with promising segmentation results, demonstrating its outperformance versus its counterparts.

Hospital Medical Behaviour Supervision and Operational Efficiency Evaluation Method based based on Big Data Platform

The application of cloud computing and big data core technologies and concepts to medical informatization can improve its flexibility and efficiency, and achieve the overall deployment and intensive management of the system. In the era of big data, the use of information management platform can optimize and standardize clinical diagnosis and treatment process, improve the quality and efficiency of diagnosis and treatment services, and improve the quality and level of scientific research, which meets the requirements of medical reform for fine management of hospitals and precise medical treatment in the era of evidence-based medicine. This paper introduces the development and application of big data analysis in the field of information technology. Through the research on the supervision of hospital medical service behavior, the top-level design is used to standardize the content of medical behavior supervision, and the effectiveness of supervision is discussed to achieve the purpose of reducing clinical paperwork. Based on the needs, the medical service behavior supervision system is proposed and constructed. Strengthen hospital medical behavior supervision through hospital big data analysis and knowledge base system support. This system can provide management with an effective monitoring and management tool, enabling them to promptly identify and solve problems that arise in medical services. By analyzing a large amount of medical data, hospitals can help predict the trend of disease occurrence in advance, thereby taking preventive and control measures in advance, and reducing the occurrence and spread of diseases. Through deep learning and analysis of patient data, personalized treatment plans can be provided for each patient to improve treatment effectiveness.

Effects of Loss Functions and Supervision Methods on Total-Body PET Denoising

Effects of Loss Functions and Supervision Methods on Total-Body PET Denoising

Learning from success in the supervision process

Supervision forms an essential component of professional social work practice. It supports the development of the capacity and skills of both students as emerging practitioners and of more experienced social workers. Reflection is a key part of supervision because social work practice is constantly evolving due to rapid change in society. Supervision provides social work practitioners and students with an opportunity to speak openly and honestly about their work, and the everyday feelings and emotions they face whilst in practice. This enables both improvement in practice and prevents the buildup of stress. In this article, we explore different methods of supervision utilised in social work and highlight implementation in Slovenia and England. We developed our own framework of supervision drawing on our knowledge of research from Slovenia and England; moreover, we engaged in informal discussion with colleagues in England, and a more formal process with supervisors in Slovenia, to inform the development of this model. This article thus reports the conceptual findings of a study and introduces our own model of reflective supervision, which is founded on the premise that we learn better from good experiences than from mistakes. We believe that through the proposed format of appreciative practice, we can move from deficit-based practice to strengths-focused practice. Key words social work; reflective practice; co-creation; group work; professional development

Investigation of Rates of Demolition Waste Generated in Decoration and Renovation Projects: An Empirical Study

There is an increase in decoration and renovation activities in the construction industry, and waste generation rates (WGRs) play a crucial role in guiding the management of demolition waste in decoration and renovation projects (DWDRPs). However, there has been little systematic research on this type of waste. Based on site surveys and a document review of 26 projects, this study offers insights into DWDRP wastes, from their initial generation to their final disposal. The results revealed that the WGRs for DWDRPs ranged from 30.96 kg/m2 to 629.96 kg/m2 and that the key components of DWDRPs included mortar, concrete, timber, tile, and metal; these five types of waste contributed 75.02% of the total waste. Although these findings deviate slightly from those of previous studies, these variations are attributed to diverse waste management practices, awareness levels, and employed construction technologies. Despite its importance, in China, the management of DWDRPs faces challenges, such as limited public awareness, inadequate collection and sorting guidance, and insufficient legislation. To counter these issues, we recommend a set of strategies, including stringent regulations, enhanced supervision, government incentives, improved collection and sorting methods, and the adoption of innovative technologies. This study not only sheds light on the specific challenges in decoration and demolition waste management in rapidly urbanizing areas but also proposes a comprehensive approach for improving waste management practices.

A medical report generation method integrating teacher–student model and encoder–decoder network

The automatic medical report generation task can reduce the burden of radiologists and improve the intelligence of auxiliary diagnosis, but still faces the following challenges: (1) The small lesions are easily overlooked, leading to loss of crucial information in the report and low accuracy; (2) The generated long text reports often suffer from jumbled word order and sentence order, resulting in poor fluency. Through simulation of the cognitive principle of professional physicians during their training and work, this paper put forward a medical report generation method integrating a teacher–student model with an encoder–decoder network. The core idea is to propose a cross-modal teacher (text)-student (image) model, adopting different supervision methods for different stages of report generation to improve the model’s learning performance. A semantic space alignment mechanism is designed to enhance the cross-modal feature matching ability by contrasting the encoding methods of different modalities through adversarial learning, gradually optimizing and capturing the critical information. A layer-supervised decoder based on the Transformer hierarchical structure is proposed with the teacher model guiding the student model to decode layer by layer to increase the fluency of report generation. Comparative experiments are conducted on IU-X-ray and MIMIC-CXR datasets with various other methods, and the results show that the proposed method can effectively improve the quality of generated reports.

Supervisees' Views on Live Observation-Based Supervision Modality

Live observation is a technique that allows the supervisor to offer immediate feedback to the supervisee by directly witnessing counseling sessions. Despite its numerous advantages over traditional supervision methods, this approach is seldom put into practice due to its time-consuming nature. Furthermore, there is insufficient research on its suitability in supervision practices, particularly in the context of Türkiye. In this study, the views of the participants on a practice in which live observation and consultation break methods used together were examined. The research involved seven students enrolled in a master's program in Guidance and Psychological Counseling at a university in the Central Black Sea region of Türkiye. Live observation and consultation break methods were incorporated into students' certain sessions as part of their individual psychological counseling practices course. After the supervision process ended, semi-structured individual interviews were conducted to evaluate the students' experiences. The data from the interviews were analyzed using content analysis. The students' opinions were grouped under four themes: development, relationship, live observation practice, and the overall supervision process. The participants stated that the live observation-based supervision practice improved their professional competence and increased their self-awareness. They also stated that they experienced anxiety at the beginning of the practice, but that this anxiety disappeared over time. The views of the participants suggest that the practice of live observation-based supervision strengthens the relationship between client, supervisee, and the supervisor. In general, the results indicate that this practice is a method that can be used in supervision processes.

Memory and relationships: A look into teachers' recollection of supervisory feedback

AbstractIn public schools in the United States, the work of an ESOL student teacher is characterized by relationships with multiple mentors, including a university supervisor. Through retrospective narrative accounts, this study examines what currently practicing teachers of multilingual learners retained of their supervisors' feedback from their student teaching days, and what nuggets of supervisory wisdom continue to resonate with them throughout their teaching careers. The researchers developed this qualitative study using a social constructivist lens informed by Carless and Winstone's (2023) feedback literacy dimensions. Interviews were conducted with eight current teachers to probe their recollections of supervision and integration of supervisory feedback into their practice. Findings suggest that the participants recalled a sprinkling of memories and continued to draw on a few pieces of supervisory advice in their current practice. Supervisors and student teachers demonstrated design, relational, and pragmatic dimensions of feedback literacy, though the relational (i.e., the quality of the relationship between supervisor and student teacher) was the most enduring, even if the specifics of the supervisory feedback faded. Implications suggest that teacher education programs can hone the purpose, mission, and methods of supervision by targeting the dimensions of feedback literacy to make supervisory feedback more constructive and impactful.

De-biased Attention Supervision for Text Classification with Causality

In text classification models, while the unsupervised attention mechanism can enhance performance, it often produces attention distributions that are puzzling to humans, such as assigning high weight to seemingly insignificant conjunctions. Recently, numerous studies have explored Attention Supervision (AS) to guide the model toward more interpretable attention distributions. However, such AS can impact classification performance, especially in specialized domains. In this paper, we address this issue from a causality perspective. Firstly, we leverage the causal graph to reveal two biases in the AS: 1) Bias caused by the label distribution of the dataset. 2) Bias caused by the words' different occurrence ranges that some words can occur across labels while others only occur in a particular label. We then propose a novel De-biased Attention Supervision (DAS) method to eliminate these biases with causal techniques. Specifically, we adopt backdoor adjustment on the label-caused bias and reduce the word-caused bias by subtracting the direct causal effect of the word. Through extensive experiments on two professional text classification datasets (e.g., medicine and law), we demonstrate that our method achieves improved classification accuracy along with more coherent attention distributions.

INTERACTIVE ROLE OF STRATEGIC CLARITY IN THE RELATIONSHIP BETWEEN ORGANIZATIONAL CONFLICT MANAGEMENT AND STRATEGIC DECISION QUALITY

Different supervision methods usually lead to internal organizational conflicts, especially in a multi-unit organization, such as The Iraqi Ministry of Interior (IMI). In this regard, Successful organizational conflict management (OCM) is fundamentally associated with Strategic Clarity (SC) in addressing such conflicts which is also linked to the Strategic Decision Quality (SDQ). However, the complexity of the abovementioned variables is understudied especially in the governmental security-service sectors. The present study provides a comprehensive analysis of the relationships between OCM, SC and SDQ in the IMI. A questionnaire, prepared for measuring the interactive role of SC in the relationship between OCM and the SDQ, was distributed among leaders and managers of that ministry. It was evident that SC significantly enhances the relationship between OCM and SDQ in the ministry, registering a substantial interactive effect of 0.25. The study concluded that the dimensions of SC amplify the strategic connection between OCM and the high-quality SDQ made by administrative leaders in the ministry. This emphasizes the need for further integration and activation of SC as part of the strategic vision, particularly in security challenges.

Exploring an application-oriented land-based hyperspectral target detection framework based on 3D–2D CNN and transfer learning

Target detection based on hyperspectral images refers to the integrated use of spatial information and spectral information to accomplish the task of localization and identification of targets. There are two main methods for hyperspectral target detection: supervised and unsupervised methods. Supervision method refers to the use of spectral differences between the target to be tested and the surrounding background to identify the target when the target spectrum is known. In ideal situations, supervised object detection algorithms perform better than unsupervised algorithms. However, the current supervised object detection algorithms mainly have two problems: firstly, the impact of uncertainty in the ground object spectrum, and secondly, the universality of the algorithm is poor. A hyperspectral target detection framework based on 3D–2D CNN and transfer learning was proposed to solve the problems of traditional supervised methods. This method first extracts multi-scale spectral information and then preprocesses hyperspectral images using multiple spectral similarity measures. This method not only extracts spectral features in advance, but also eliminates the influence of complex environments to a certain extent. The preprocessed feature maps are used as input for 3D–2D CNN to deeply learn the features of the target, and then, the softmax method is used to output and obtain the detection results. The framework draws on the ideas of integrated learning and transfer learning, solves the spectral uncertainty problem with the combined similarity measure and depth feature extraction network, and solves the problem of poor robustness of traditional algorithms by model migration and parameter sharing. The area under the ROC curve of the proposed method has been increased to over 0.99 in experiments on both publicly available remote sensing hyperspectral images and measured land-based hyperspectral images. The availability and stability of the proposed method have been demonstrated through experiments. A feasible approach has been provided for the development and application of specific target detection technology in hyperspectral images under different backgrounds in the future.