Iris verification systems provide reliable biometric authentication, owing to the distinctiveness and lifelong stability of iris patterns. Modern pipelines such as Open-IRIS extend Daugman’s classical framework with a Gabor filter bank, where each filter response is encoded into a binary iris code. Consequently, iterating over per-filter iris codes at every cyclic shift causes per-comparison matching cost to scale linearly with the number of filters, yielding a measured total matching time of 5.38 min on the CASIA-IrisV2 dataset. This work introduces Template Vote Fusion, a consolidation mechanism that retains only bits confirmed by majority voting on phase texture bits and consensus masking on occlusion masks, supported by a mid-frequency filter derived as the arithmetic mean of the existing filter parameters to ensure three voters. Template Vote Fusion reduces full-pipeline latency from 2.024 ms to 0.803 ms and per-comparison matching time from 1.743 ms to 0.543 ms, while the equal error rate decreases from 0.1409
This study investigates the determinants of agricultural and livestock production across 23 Asian countries from 2000 to 2023, with a particular emphasis on climate-related risks. Using a Panel ARDL model, the analysis distinguishes between short-run adjustments and long-run equilibrium effects associated with physical and transition climate risks. The results indicate that the Climate Physical Risk Index (CPRI) negatively affects both crop and livestock production in the short run, but its long-run impact is positive, reflecting gradual adaptation through improved farming practices and resilience-building strategies. In contrast, renewable energy—representing transition risk—has a positive short-run effect on crop production yet exerts significant negative long-run impacts on both sectors, likely due to land competition, transition costs, and energy infrastructure constraints. Furthermore, the Temperature-Humidity Index (THI) imposes substantial long-run negative effects on livestock productivity due to chronic heat stress and associated physiological pressures. These findings highlight the complex and time-varying pathways through which climate risks influence agricultural systems and underscore the need for policies that enhance adaptive capacity, safeguard food security, and support climate-resilient development.
Software Defect Prediction (SDP) aims to identify defect-prone modules using historical software metrics. However, conventional predictive models often face limitations due to ineffective feature representation and class imbalance. This paper proposes an effective framework based on Intuitionistic Fuzzy Set (IFS) theory combined with an adaptive information transformation operator. The methodology consists of two stages: (i) fuzzification of features into IFS and (ii) classification using an Adaptive Intuitionistic Fuzzy K-Nearest Neighbors (Adaptive IF-KNN) algorithm. A key aspect of this approach is an adaptive weighting mechanism that redistributes hesitation degrees according to the class imbalance ratio. Experimental results on five benchmark datasets from NASA and PROMISE demonstrate the superiority of the proposed framework in addressing the inherent challenges of SDP.
This study aims to identify structural imbalances within professional and educational teams by proposing the relative efficiency variance index (REVI). While conventional performance evaluations often focus narrowly on low-performers in a team, our approach prioritizes team harmony and the long-term sustainability of collaborative activities by focusing on internal balance. Utilizing interval efficiency analysis from both optimistic and pessimistic perspectives, the REVI quantifies the evaluative overlap and the concentration of low-performing members to detect unhealthy disparities.
This paper proposes a sector-oriented data science framework for assessing village development and Sustainable Development Goals (SDGs)-related conditions using Village Potential Statistics (PODES) data from 503 villages in Bogor, West Java, Indonesia. The dataset contains mixed numerical and categorical variables organized into eight sectors. Since no single global village development index is available, the analysis is formulated as sector-specific classification and regression tasks with policy-relevant targets. The framework integrates data cleaning, exploratory analysis, target screening, feature selection, comparative model evaluation, and visualization. Experimental results show that classification achieves generally moderate to strong performance across sectors, while six retained regression sectors obtain positive R^2 values, with the strongest results in health and education, natural disaster and environment, and infrastructure. The results suggest that village development patterns vary substantially across sectors and that sector-oriented target design, combined with appropriate sector-wise and cross-sector feature selection, provides a practical basis for village-level development assessment.
