Abstract

Assessing financial resilience in the banking sector requires an integrated framework that captures both cross-sectional strength and long-term resilience dynamics. In this study, financial resilience of Indian commercial banks during the period 2013–2024 is assessed by using a hybrid MCDM and non-parametric trend analysis approach. Eleven financial indicators covering solvency, asset quality, efficiency, and profitability are included in a composite resilience framework, and criterion weights were objectively determined using the MEREC method. The technique RAM is used to calculate annual composite resilience scores and ranks the 29 commercial banks. A Mann–Kendall time-series analysis is also applied to the final RAM scores to analyze long-term monotonic trends in bank-level and sector-wide resilience. The results showed that RAM scores are tightly clustered across banks, suggesting structural convergence in resilience levels. However, Kruskal-Wallis non-parametric test showed statistically significant differences in the banks’ relative financial resilience across the study period. The MEREC-RAM ranking result showed Kotak Mahindra Bank Ltd. and Tamilnad Mercantile Bank Ltd. consistently appeared among top at the rankings. While, the Mann-Kendall trend test revealed significant improvement in the resilience of CSB Bank Ltd. and Bank of Maharashtra over the study period. Overall, combining year-wise relative rankings and monotonic resilience dynamics enables a comprehensive assessment of the stability of the Indian banking sector, which can offer key insights for regulators, policymakers, and bank management in strengthening the long-term financial resilience of the sector.

Assessing financial resilience in the banking sector requires an integrated framework that captures both cross-sectional strength and long-term resilience dynamics. In this study, financial resilience of Indian commercial banks during the period 2013–2024 is assessed by using a hybrid MCDM and non-parametric trend analysis approach. Eleven financial indicators covering solvency, asset quality, efficiency, and profitability are included in a composite resilience framework, and criterion weights were objectively determined using the MEREC method. The technique RAM is used to calculate annual composite resilience scores and ranks the 29 commercial banks. A Mann–Kendall time-series analysis is also applied to the final RAM scores to analyze long-term monotonic trends in bank-level and sector-wide resilience. The results showed that RAM scores are tightly clustered across banks, suggesting structural convergence in resilience levels. However, Kruskal-Wallis non-parametric test showed statistically significant differences in the banks’ relative financial resilience across the study period. The MEREC-RAM ranking result showed Kotak Mahindra Bank Ltd. and Tamilnad Mercantile Bank Ltd. consistently appeared among top at the rankings. While, the Mann-Kendall trend test revealed significant improvement in the resilience of CSB Bank Ltd. and Bank of Maharashtra over the study period. Overall, combining year-wise relative rankings and monotonic resilience dynamics enables a comprehensive assessment of the stability of the Indian banking sector, which can offer key insights for regulators, policymakers, and bank management in strengthening the long-term financial resilience of the sector.

Abstract

Commonsense is a broad and multifaceted concept, making its evaluation a persistent challenge in natural language processing (NLP). This paper introduces JAMAL (Arabic for “Camel”), a multidimensional framework and benchmark for Arabic commonsense reasoning. JAMAL is structured along three complementary axes: (i) a life-domain axis comprising a taxonomy of 56 functional categories informed by the World Health Organization’s International Classification of Functioning, Disability, and Health (ICF), capturing diverse aspects of daily human experience; (ii) a cognitive axis organizing commonsense into three reasoning types: everyday situations, general knowledge, and problem-solving; and (iii) a cultural grounding axis distinguishing between universal, western/global, and Arabic-specific knowledge. To operationalize this framework, benchmark items are constructed using psycholinguistically inspired principles of constrained contextual prediction. We evaluate five Arabic language models using JAMAL and observe consistent differences in their performance across all axes. Notably, FANAR-27B achieves the strongest overall results among all evaluated models, outperforming FANAR-9B and smaller baselines. Overall, JAMAL provides a structured and interpretable benchmark for evaluating commonsense reasoning in Arabic, supporting the development of more robust language models through systematic analysis of their behavioral limitations.

Commonsense is a broad and multifaceted concept, making its evaluation a persistent challenge in natural language processing (NLP). This paper introduces JAMAL (Arabic for “Camel”), a multidimensional framework and benchmark for Arabic commonsense reasoning. JAMAL is structured along three complementary axes: (i) a life-domain axis comprising a taxonomy of 56 functional categories informed by the World Health Organization’s International Classification of Functioning, Disability, and Health (ICF), capturing diverse aspects of daily human experience; (ii) a cognitive axis organizing commonsense into three reasoning types: everyday situations, general knowledge, and problem-solving; and (iii) a cultural grounding axis distinguishing between universal, western/global, and Arabic-specific knowledge. To operationalize this framework, benchmark items are constructed using psycholinguistically inspired principles of constrained contextual prediction. We evaluate five Arabic language models using JAMAL and observe consistent differences in their performance across all axes. Notably, FANAR-27B achieves the strongest overall results among all evaluated models, outperforming FANAR-9B and smaller baselines. Overall, JAMAL provides a structured and interpretable benchmark for evaluating commonsense reasoning in Arabic, supporting the development of more robust language models through systematic analysis of their behavioral limitations.

Abstract

Alzheimer’s disease is a progressive neurodegenerative disorder that significantly impairs memory and cognitive functions and affects over 55 million people worldwide. The successful management and planning require early and accurate diagnosis. Conventional radiological assessment is often subjective and time-consuming, which highlights the need for automated and reliable diagnostic solutions. Most deep learning models show promise for classifying neuroimaging data, but they tend to be less computationally efficient and less interpretable, and they cannot be integrated into patient-centric processes. The gap between developing diagnostic algorithms with high accuracy and implementing them in a supportive framework that includes patients and caregivers is large. This paper introduces a comprehensive, hybrid framework that addresses these gaps. We present a dual-modality diagnostic system: a deep learning pipeline using EfficientNetV2-S for CT scan classification, complemented by a Feedforward Neural Network (FNN) that analyses structured clinical data for holistic patient assessment. This diagnostic core is integrated into a user-friendly graphical user interface (GUI) and supplemented by ”NeuroBot,” an AI-powered chatbot that provides domain-specific information and support. The two models have been trained using the transfer learning method on a curated dataset of 30,000 brain CT slices. The EfficientNetV2-S model achieved an accuracy of 98.19%. After hyperparameter tuning, the FNN model achieved an optimised accuracy of 87.21%. The importance of the features addressed by the models was proved with the help of the statistical t-tests of the corresponding clinical data. The integrated system enables a scalable, translatable, and patient-centered system to improve the early analysis and treatment of Alzheimer’s disease.

Alzheimer’s disease is a progressive neurodegenerative disorder that significantly impairs memory and cognitive functions and affects over 55 million people worldwide. The successful management and planning require early and accurate diagnosis. Conventional radiological assessment is often subjective and time-consuming, which highlights the need for automated and reliable diagnostic solutions. Most deep learning models show promise for classifying neuroimaging data, but they tend to be less computationally efficient and less interpretable, and they cannot be integrated into patient-centric processes. The gap between developing diagnostic algorithms with high accuracy and implementing them in a supportive framework that includes patients and caregivers is large. This paper introduces a comprehensive, hybrid framework that addresses these gaps. We present a dual-modality diagnostic system: a deep learning pipeline using EfficientNetV2-S for CT scan classification, complemented by a Feedforward Neural Network (FNN) that analyses structured clinical data for holistic patient assessment. This diagnostic core is integrated into a user-friendly graphical user interface (GUI) and supplemented by ”NeuroBot,” an AI-powered chatbot that provides domain-specific information and support. The two models have been trained using the transfer learning method on a curated dataset of 30,000 brain CT slices. The EfficientNetV2-S model achieved an accuracy of 98.19%. After hyperparameter tuning, the FNN model achieved an optimised accuracy of 87.21%. The importance of the features addressed by the models was proved with the help of the statistical t-tests of the corresponding clinical data. The integrated system enables a scalable, translatable, and patient-centered system to improve the early analysis and treatment of Alzheimer’s disease.