Search Results (72)

This study explores the challenges and theoretical transformations that the widespread application of AI technology in social governance brings to the protection of citizens’ fundamental rights. By examining typical cases in judicial assistance, technology-enabled law enforcement, and welfare supervision, it explains how AI characteristics such as algorithmic opacity, data bias, and automated decision-making affect fundamental rights including due process, equal protection, and privacy. The article traces the historical evolution of privacy theory from physical space protection to informational self-determination and further to modern data rights, pointing out the inadequacy of traditional rights-protection paradigms in addressing the characteristics of AI technology. Through analyzing AI-governance models in the European Union, the United States, Northeast Asia, and international organizations, it demonstrates diverse governance approaches ranging from systematic risk regulation to decentralized industry regulation. With a special focus on China, the article analyzes the special challenges faced in AI governance and proposes specific recommendations for improving AI-governance paths. The article argues that only within the track of the rule of law, through continuous theoretical innovation, institutional construction, and international cooperation, can AI technology development be ensured to serve human dignity, freedom, and fair justice. Full article

With the rapid development of artificial intelligence technology, DRL has shown great potential in solving complex real-time optimal power flow problems of modern power systems. Nevertheless, traditional DRL methodologies confront dual bottlenecks: (a) suboptimal coordination between exploratory behavior policies and experience-based data exploitation in practical applications, compounded by (b) users’ distrust from the opacity of model decision mechanics. To address these, a model–data hybrid-driven physics-informed reinforcement learning (PIRL) algorithm is proposed in this paper. Specifically, the proposed methodology uses the proximal policy optimization (PPO) algorithm as the agent’s foundational framework and constructs a PI-actor network embedded with prior model knowledge derived from power flow sensitivity into the agent’s actor network via the PINN method, which achieves dual optimization objectives: (a) enhanced environmental perceptibility to improve experience utilization efficiency via gradient-awareness from model knowledge during actor network updates, and (b) improved user trustworthiness through mathematically constrained action gradient information derived from explicit model knowledge, ensuring actor updates adhere to safety boundaries. The simulation and validation results show that the PIRL algorithm outperforms the baseline PPO algorithm in terms of training stability, exploration efficiency, economy, and security. Full article

Pectin–tomato paste edible films with potential antioxidant activity were studied. Initially, the films were formed by drying at 40 °C in the presence and absence of glycerol. The effect of drying temperature on several physicochemical, mechanical, and optical properties of glycerol films formed after drying at 40, 50, and 60 °C was investigated. Finally, films formed at different drying conditions (namely F40, F50, and F60) sharing the same antioxidant activity (44.28–45.53%) were studied in terms of their surface pH; solubility; folding endurance; antimicrobial, dynamic mechanical, and barrier properties; contact angle; and FT-IR. Their thickness, weight, opacity, strength, stiffness, and antioxidant activity (AA) [a*] increased with increasing tomato paste content, whereas [L*] decreased. The moisture content was statistically affected by both the presence of glycerol and the drying temperature. AA decreased as drying temperature increased. Overall, the thickness varied from 45 to 182.31 μm, weight from 0.27 to 1.24 g, moisture content from 20.74 to 56.66%, stress from 189 to 959 kPa, Young’s modulus from 86 to 382 kPa, and AA from 16.9 to 53%. In the last step, F60 was less hydrophilic, had a greater density, and better barrier properties, whereas F50 was stiffer and the least strong. Our findings provide information regarding the selection of an optimum drying temperature for pectin-based films with antioxidant properties. Full article

This study investigates whether and how the complexity of derivative use influences the stock price crash risk in China’s capital market, a critical question given the growing use of derivatives in emerging economies where governance structures and disclosure standards vary widely. While prior research has examined the binary effects of derivative usage, limited attention has been paid to the multidimensional complexity of such instruments and its informational consequences. Using a novel hand-collected dataset of annual reports from Chinese A-share-listed firms between 2010 and 2023, we develop and implement new indicators that capture both the economic complexity (diversity and scale) and accounting complexity (reporting dispersion and fair-value hierarchy) of derivative use. Our analysis shows that higher complexity is associated with a significantly lower likelihood of stock price crashes. This effect is especially pronounced in non-state-owned firms and those with weaker internal-control systems, suggesting that derivative complexity can enhance information transparency and serve as a substitute for other governance mechanisms. These findings challenge the conventional view that complexity necessarily increases opacity and highlight the importance of disclosure quality and institutional context in shaping the market consequences of financial innovation. Full article

Molar Incisor Hypomineralization (MIH) is a common dental condition that affects the mineralization of the enamel, primarily affecting the first permanent molars and often the incisors. This condition can lead to a wide range of clinical presentations, from mild opacities to severe post-eruptive breakdown, which can significantly impact a child’s oral health and quality of life. Background/Objectives: The prevalence and complex management of MIH have posed a significant challenge for dental practitioners. Our preceding investigation found that 14.3% of school-aged children have MIH. Based on this finding, we wanted to understand what other Romanian dental professionals think about this condition. Therefore, the aim of the present study was to assess the awareness, perception, and clinical management approaches of Romanian dentists toward MIH in order to inform future educational strategies and contribute to the development of dedicated preventive programs. Methods: To gain a comprehensive understanding of MIH in actual clinical settings, we developed and administered a questionnaire consisting of three distinct sections. Our objective was to capture the collective knowledge and perspectives of dental practitioners. We distributed the survey, which included 14 pertinent questions, to a large professional group of Romanian dentists. Results: This study collected responses from 219 Romanian dental practitioners (median age: 34 years) about their experiences with MIH. The vast majority (86.76%) had encountered MIH cases in their practice, with half reporting moderate prevalence among their patients. The most frequently observed complications were hypersensitivity (41.95%), pulp exposure (33.33%), and failed restorations (24.71%). While adhesive restorations were identified as the overall preferred treatment approach (70.00%), notable differences emerged in both clinical complications encountered and therapeutic approaches implemented across dental specialties. There was near-unanimous agreement on the importance of early MIH diagnosis (99.09%), and almost all participants (98.63%) expressed a desire for more information about this condition, demonstrating high awareness and concern about MIH among Romanian dental professionals. Conclusions: This study highlights that general dentists, endodontists, and pedodontists encounter MIH patients frequently in their practice, emphasizing the critical need to enhance awareness and education about MIH among both dental professionals and the general public. Full article

This article presents a novel methodology for auditing urban regulatory frameworks through the application of artificial intelligence (AI) using the case of Greater Santiago as an empirical laboratory. Based on the semantic analysis of 31 communal zoning ordinances (Planes Reguladores Comunales, PRCs), the study uncovers how legal structures actively reproduce socio-spatial inequalities under the guise of normative neutrality. The DeepSeek-R1 model, fine-tuned for Chilean legal-urban discourse, was used, enabling the detection of normative asymmetries, omissions, and structural fragmentation. Key findings indicate that affluent communes, such as Vitacura and Las Condes, display detailed and incentive-rich regulations, while peripheral municipalities lack provisions for social housing, participatory mechanisms, or climate resilience, thereby reinforcing exclusionary patterns. The analysis also introduces a scalable rubric-based evaluation system and GIS visualizations to synthetize regulatory disparities across the metropolitan area. Methodologically, the study shows how domain-adapted AI can extend regulatory scrutiny beyond manual limitations, while substantively contributing to debates on spatial justice, institutional fragmentation, and regulatory opacity in urban planning. The results call for binding mechanisms that align local zoning with metropolitan equity goals and highlight the potential of automated audits to inform reform agendas in the Global South. Full article

Opacity, an important property of the information flow in discrete-event systems (DESs), characterizes whether the secret information in a system is ambiguous to a passive observer (called an intruder). Observation models play a critical role in the analysis of opacity. In this paper, instead of adopting a fully static observation model or a fully dynamic observation model, we use a novel Orwellian-type observation model to study the verification of the current-state opacity (CSO), where the observability of an unobservable event can be re-interpreted once certain/several specific conditions are met. First, a K-delay Orwellian observation mechanism (KOOM) is proposed as a novel Orwellian-type observation mechanism for extending the existing Orwellian projection. The main characteristics of the KOOM are delaying the inevitable information release and narrowing the release range for historical information to protect the secrets in a system to a greater extent than with the existing Orwellian projection. Second, we formulate the definitions of standard and strong CSO under the KOOM. Finally, we address the verification problem for these two types of opacity by constructing two novel information structures called a standard K-delay verifier and a strong K-delay verifier, respectively. An analysis of the computational complexity and illustrative examples are also presented for the proposed results. Overall, the proposed notions of standard and strong CSO under the KOOM capture the security privacy requirements regarding a delayed release in applications, such as intelligent transportation systems, etc. Full article

Despite their unprecedented success, artificial neural networks suffer extreme opacity and weakness in learning general knowledge from limited experience. Some argue that the key to overcoming those limitations in artificial neural networks is efficiently combining continuity with compositionality principles. While it is unknown how the brain encodes and decodes information in a way that enables both rapid responses and complex processing, there is evidence that the neocortex employs sparse distributed representations for this task. This is an active area of research. This work deals with one of the challenges in this field related to encoding and decoding nested compositional structures, which are essential for representing complex real-world concepts. One of the algorithms in this field is called context-dependent thinning (CDT). A distinguishing feature of CDT relative to other methods is that the CDT-encoded vector remains similar to each component input and combinations of similar inputs. In this work, we propose a novel encoding method termed CPSE, based on CDT ideas. In addition, we propose a novel decoding method termed CPSD, based on triadic memory. The proposed algorithms extend CDT by allowing both encoding and decoding of information, including the composition order. In addition, the proposed algorithms allow to optimize the amount of compute and memory needed to achieve the desired encoding/decoding performance. Full article

This paper delves into current-state opacity enforcement in partially observed discrete event systems through an innovative application of differential privacy, which is fundamental for security-critical cyber–physical systems. An opaque system implies that an external agent cannot infer the predefined system secret via its observational output, such that the important system information flow cannot be leaked out. Differential privacy emerges as a robust framework that is pivotal for the protection of individual data integrity within these systems. Motivated by the differential privacy mechanism for information protection, this research proposes the secret string adjacency relation as a novel concept, assessing the similarity between potentially compromised strings and system-generated alternatives, thereby shielding the system’s confidential data from external observation. The development of secret string differential privacy is achieved by substituting sensitive strings. These substitution strings are generated by a modified Levenshtein automaton, following exponentially distributed generation probabilities. The verification and illustrative examples of the proposed mechanism are provided. Full article

The use of artificial intelligence (AI) in journalism has transformed the sector, with media generating content automatically without journalists’ involvement, and various media companies implementing AI solutions. Some research suggests AI-authored articles are perceived as equally credible as human-written content, while others raise concerns about misinformation and trust erosion Most studies focus on journalists’ views, with audience attitudes explored mainly through quantitative methods, though there is no consensus regarding the acceptability of AI use by news organizations. We explore AI’s role in journalism through audience research, conducting five focus groups to understand public perceptions. The findings highlight concerns about AI-generated content, particularly potential errors, opacity, and coldness of the content. The information is perceived as somewhat less valuable, being viewed as more automated and requiring less human effort. These concerns coexist with a certain view of AI content as more objective, unbiased, and closer to the ideal of independence from political and economic pressures. Nevertheless, citizens with more AI knowledge question the neutrality of automated content, suspecting biases from corporate interests or journalists influencing the prompts. Full article

Objectives: Prediction of lung function deficits following pulmonary infection is challenging and suffers from inaccuracy. We sought to develop machine-learning models for prediction of post-inflammatory lung changes based on COVID-19 recovery data. Methods: In the prospective CovILD study (n = 420 longitudinal observations from n = 140 COVID-19 survivors), data on lung function testing (LFT), chest CT including severity scoring by a human radiologist and density measurement by artificial intelligence, demography, and persistent symptoms were collected. This information was used to develop models of numeric readouts and abnormalities of LFT with four machine learning algorithms (Random Forest, gradient boosted machines, neural network, and support vector machines). Results: Reduced DLCO (diffusion capacity for carbon monoxide <80% of reference) was found in 94 (22%) observations. Those observations were modeled with a cross-validated accuracy of 82–85%, AUC of 0.87–0.9, and Cohen’s κ of 0.45–0.5. No reliable models could be established for FEV1 or FVC. For DLCO as a continuous variable, three machine learning algorithms yielded meaningful models with cross-validated mean absolute errors of 11.6–12.5% and R² of 0.26–0.34. CT-derived features such as opacity, high opacity, and CT severity score were among the most influential predictors of DLCO impairment. Conclusions: Multi-parameter machine learning trained with demographic, clinical, and artificial intelligence chest CT data reliably and reproducibly predicts LFT deficits and outperforms single markers of lung pathology and human radiologist’s assessment. It may improve diagnostic and foster personalized treatment. Full article

With the advancements in deep learning methods, AI systems now perform at the same or higher level than human intelligence in many complex real-world problems. The data and algorithmic opacity of deep learning models, however, make the task of comprehending the input data information, the model, and model’s decisions quite challenging. This lack of transparency constitutes both a practical and an ethical issue. For the present study, it is a major drawback to the deployment of deep learning methods mandated with detecting patterns and prognosticating Alzheimer’s disease. Many approaches presented in the AI and medical literature for overcoming this critical weakness are sometimes at the cost of sacrificing accuracy for interpretability. This study is an attempt at addressing this challenge and fostering transparency and reliability in AI-driven healthcare solutions. The study explores a few commonly used perturbation-based interpretability (LIME) and gradient-based interpretability (Saliency and Grad-CAM) approaches for visualizing and explaining the dataset, models, and decisions of MRI image-based Alzheimer’s disease identification using the diagnostic and predictive strengths of an ensemble framework comprising Convolutional Neural Networks (CNNs) architectures (Custom multi-classifier CNN, VGG-19, ResNet, MobileNet, EfficientNet, DenseNet), and a Vision Transformer (ViT). The experimental results show the stacking ensemble achieving a remarkable accuracy of 98.0% while the hard voting ensemble reached 97.0%. The findings present a valuable contribution to the growing field of explainable artificial intelligence (XAI) in medical imaging, helping end users and researchers to gain deep understanding of the backstory behind medical image dataset and deep learning model’s decisions. Full article

As an important pillar of the national economy, the green transformation of SMEs is the key to promoting sustainable economic development. However, SMEs generally face issues such as information opacity and high operational risks, which make it difficult for them to obtain traditional financing support, thereby hindering green development. Green Supply Chain Finance has opened up new financing channels for SMEs, but the accuracy of credit risk evaluation remains a bottleneck that limits its widespread application. This paper constructs a credit risk evaluation index system that integrates multiple sources of information, covering factors such as the situations of SMEs themselves, stakeholder feedback, and expert ratings. It compares and analyzes the performance of the genetic algorithm-optimized random forest model (GA-RF), the BP neural network, the support vector machine, and the logistic regression model in credit risk evaluation. The empirical results indicate that the GA-RF model is significantly better than the other models in terms of accuracy, precision, and F1 score, and has the highest AUC value, making it more effective in identifying credit risk. In addition, the GA-RF model reveals that the asset–liability ratio, the time of establishment, the growth rate of operating revenue, the time of collection of accounts receivable, the return on net assets, and daily shipments are the key indicators affecting the credit risk assessment. Full article

This paper proposes a method for visualizing three-dimensional non-uniformly sampled data from multibeam echosounder systems (MBESs), aimed at addressing the requirements of monitoring complex and dynamic underwater flow fields. To tackle the challenges associated with spatially non-uniform sampling, the proposed method employs linear interpolation along the radial direction and arc length weighted interpolation in the beam direction. This approach ensures consistent resolution of three-dimensional data across the same dimension. Additionally, an opacity transfer function is generated to enhance the visualization performance of the ray casting algorithm. This function leverages data values and gradient information, including the first and second directional derivatives, to suppress the rendering of background and non-interest regions while emphasizing target areas and boundary features. The simulation and experimental results demonstrate that, compared to conventional two-dimensional beam images and three-dimensional images, the proposed algorithm provides a more intuitive and accurate representation of three-dimensional data, offering significant support for the observation and analysis of spatial flow field characteristics. Full article

Background and Objectives: Thin-section CT (TSCT) is currently the most sensitive imaging modality for detecting bronchiectasis. However, conventional TSCT or HRCT may overlook subtle lung involvement such as alveolar and interstitial changes. Artificial Intelligence (AI)-based analysis offers the potential to identify novel information on lung parenchymal involvement that is not easily detectable with traditional imaging techniques. This study aimed to assess lung involvement in patients with bronchiectasis using the Bronchiectasis Radiologically Indexed CT Score (BRICS) and AI-based quantitative lung texture analysis software (IMBIO, Version 2.2.0). Methods: A cross-sectional study was conducted on 45 subjects diagnosed with bronchiectasis. The BRICS severity score was used to classify the severity of bronchiectasis into four categories: Mild, Moderate, Severe, and tractional bronchiectasis. Lung texture mapping using the IMBIO AI software tool was performed to identify abnormal lung textures, specifically focusing on detecting alveolar and interstitial involvement. Results: Based on the Bronchiectasis Radiologically Indexed CT Score (BRICS), the severity of bronchiectasis was classified as Mild in 4 (8.9%) participants, Moderate in 14 (31.1%), Severe in 11 (24.4%), and tractional in 16 (35.6%). AI-based lung texture analysis using IMBIO identified significant alveolar and interstitial abnormalities, offering insights beyond conventional HRCT findings. This study revealed trends in lung hyperlucency, ground-glass opacity, reticular changes, and honeycombing across severity levels, with advanced disease stages showing more pronounced structural and vascular alterations. Elevated pulmonary vascular volume (PVV) was noted in cases with higher BRICSs, suggesting increased vascular remodeling in severe and tractional types. Conclusions: AI-based lung texture analysis provides valuable insights into lung parenchymal involvement in bronchiectasis that may not be detectable through conventional HRCT. Identifying significant alveolar and interstitial abnormalities underscores the potential impact of AI on improving the understanding of disease pathology and disease progression, and guiding future therapeutic strategies. Full article