Although machine learning-based student dropout prediction models achieve high predictive performance, the socioeconomic biases and decision-making mechanisms latent within these black-box models remain insufficiently understood. Conventional eXplainable artificial intelligence (XAI) techniques rely on correlation-based global feature importance, which limits their ability to identify the true causal factors that substantively flip predictions. Therefore, this study proposes a "Causal Auditing" framework to quantitatively identify and correct biases. Utilizing a Portuguese university student dataset and a single ensemble baseline model (LightGBM), we generated hierarchical multivariate counterfactual scenarios that combined uncontrollable sensitive attributes (parents' background) with actionable variables (academic achievement and debt status). The results revealed a distinct "Causal Mismatch" phenomenon: although the correlation-based SHapley Additive exPlanations analysis assigned the highest importance to past first-semester grades, the core causal triggers that flipped the model decision from "Dropout" to "Graduate" were found to be second-semester curricular units and financial status. Notably, instead of forcing extreme high-risk students to achieve unreasonable grade improvements, merely resolving their debt issues and neutralizing the structural penalties associated with their parental backgrounds—when combined with minimal academic support (completing +1 additional course)—successfully flipped the predictions for the entire target cohort. These findings demonstrate that AI models adopt students' environmental disadvantages as a form of proxy bias (i.e., a "structural penalty”). The proposed framework provides a systematic methodology for verifying the transparency and causal fairness of AI in the educational domain, and contributes to the formulation of data-driven, hybrid student support policies in university administrations.
Objective: The goal of this research is to use scalp and hair images to create an efficient deep learning framework for the automated detection and severity categorization of Alopecia Areata (AA).Methods: Using scalp and hair images collected from the Derm-Net and Figaro1K databases, the proposed ADAR-Net framework is developed to detect and assess Alopecia Areata. Important local and global characteristics are first extracted from the input images by two distinct FRCNN-based encoder networks. The model combines Cross-Residual Learning (CRL) and Multi-Task Learning (MTL) strategies to improve feature learning and enhance information transfer between related tasks. The relationships between the collected characteristics are then captured by combining them and passing them via an LSTM layer. However, the model is able to focus on the most important areas of the images while minimizing the impact of irrelevant data thanks to Channel Attention and Spatial Attention methods. The images are then classified into four categories: Healthy Alopecia Areata, Mild Alopecia Areata, Moderate Alopecia Areata, and Severe Alopecia Areata after the improved features are put into a fully connected layer and a SoftMax classifier.Conclusion: The test findings demonstrate that, in comparison to all previous conventional systems, the proposed Attention-Driven Alopecia detection Network (ADAR-Net), which is designed to enhance the automated detection and classification of Alopecia Areata using hair and scalp image databases, achieves an accuracy of 95.1%.Novelty: The combination of the CRL, MTL, LSTM, and attention processes in a single model for Alopecia Areata identification is what makes this work innovative. By using attention-guided learning, the suggested architecture concurrently learns complementing scalp and hair characteristics and improves their representation.
Artificial Intelligence (AI) has emerged as a critical driver of transformation in the retail sector, enabling data-driven decision-making and enhanced operational efficiency. This study examines the impact of AI adoption on decision-making efficiency and retail performance in the Indian retail sector. A sample of 500 retail firms across major urban centers was analyzed using advanced statistical techniques, including descriptive statistics, reliability analysis, correlation, exploratory factor analysis (EFA), multiple regression, and structural equation modeling (SEM). The findings reveal that AI adoption significantly improves decision efficiency and customer analytics, which in turn enhance overall firm performance. The study contributes to the literature by providing empirical evidence from an emerging economy and offers practical implications for retail managers seeking competitive advantage through AI.
Background: The high growth in digital banking has led to more demand in the use of enhanced, secure, and user-friendly authentication. Iris recognition is a next-generation biometric technology that is highly accurate, does not spoof, and is more convenient to use. AI-enabled iris recognition can enhance biometric authentication by applying intelligent pattern recognition to iris features, supporting more accurate identity verification and stronger protection against fraudulent access in digital banking environments. Nonetheless, there is little empirical evidence regarding its acceptance and perceived security and influence on digital banking security. These AI-enabled biometric capabilities are particularly relevant to fraud prevention and financial security because they can strengthen identity verification and reduce opportunities for unauthorized access to digital banking services.Objectives: This paper explores the effectiveness of technological features of iris recognition in digital banking, such as accuracy, anti-spoofing, system response time, and user interface experience in influencing the security of digital banking as well as user adoption. It also looks into the mediating role of the perceived security as well as the moderating role of trust in biometric technologies. The study also considers the relevance of AI-enabled iris recognition to fraud prevention and financial security by examining whether enhanced biometric authentication can strengthen digital banking security and reduce the risk of unauthorized access.Methods: The quantitative research design was used, and the data were gathered using 233 respondents (digital banking users) through a structured Likert-scale questionnaire. The Shapiro-Wilk, Cronbach's Alpha, KMO, and Bartlett tests were used to confirm normality, reliability, as well as validity. Independent samples t-test, One-way ANOVA, Chi-square, and Kruskal-Wallis tests were used to analyze group differences. Pearson correlation and multiple regression were the two methods employed to test the relationship between variables. The questionnaire therefore assessed the perceived effectiveness of AI-enabled iris recognition, including biometric accuracy, anti-spoofing capability, security, and authentication performance in relation to fraud prevention and financial security.Results: The reliability of all the instruments was excellent (α ≥ 0.70), and the validity (KMO = 0.842). Results showed there were strong and positive relationships between all variables. Findings of the regression tests proved the existence of a significant correlation between accuracy, anti-spoofing technology, system response time, user experience, perceived security, and trust in increasing user adoption of the iris recognition. The strongest predictor was perceived security. The difference tests of the group presented the results that gender, age, education level, and biometric experience had a significant effect on security perceptions and adoption intentions. The findings further indicate that AI-enabled iris recognition capabilities, particularly accuracy, anti-spoofing performance, and perceived security, are positively associated with stronger digital banking protection and fraud-prevention potential.Conclusion: The paper has come to the conclusion that iris recognition is a highly potent and efficient biometric remedy to improving the security of digital banking. High perceived security and trust in conjunction with technological performance create a great user adoption. With the application of awareness and trust-building techniques and user-centric design, iris recognition has high potential of becoming a mainstream authentication in financial services. The findings further support the potential of AI-enabled iris recognition as an intelligent biometric security mechanism for fraud prevention and strengthening financial security in digital banking.
With the widespread use of social media platforms, there is a massive amount of user-generated text which necessitates performing sentiment analysis, a natural language processing task. Nevertheless, correctly identifying the sentiment proves to be a difficult task because of the occurrence of sarcasm where the literal sentiment is opposite to the intended meaning. Prior research has either addressed sentiment analysis and sarcasm detection as separate tasks or developed static ensemble methods that do not exploit the semantic relationship between the two tasks. To overcome these drawbacks, this work presents AMTF-Net (Adaptive Multi-Task Transformer Fusion Network), a unified multi-task learning framework capable of jointly performing sentiment analysis and sarcasm detection. The framework extracts complementary contextual representations using BERT, GPT-2, and XLNet. The extracted contextual representations are fused through a MATF module that adaptively learns the contribution of each transformer for every input. The Shared Semantic Encoder (SSE) and Cross-Task Attention Interaction (CTAI) module refine the fused features to allow knowledge sharing between the two tasks. A Dynamic Joint Loss (DJL) function is used, which adaptively balances the optimization objectives of both tasks. The three public benchmark datasets employed were Twitter, Reddit, and HuffPost news headlines. The experimental results show that AMTF-Net has superior performance against individual transformers, conventional ensemble methods, and recent state-of-the-art methods. The proposed framework was evaluated on Twitter, Reddit, and HuffPost benchmark datasets for joint sentiment analysis and sarcasm detection. The proposed framework achieved an accuracy of 97.74 %, 98.12 % and 98.05 % respectively. Also, the Macro F1-scores were achieved at 98.12 %, 98.04 % and 97.82 % respectively for each dataset for sarcasm detection. These findings show that adaptive transformer fusion and cross-task semantic interaction enhance the robustness and accuracy of joint sentiment analysis and sarcasm detection.
