833 papers
This paper introduces KinderMM-Cap, a specialized multimedia captioning system for early childhood education that leverages the ECAC benchmark and a novel reward-conditional hybrid training framework (RSRS) to significantly improve the accuracy of generating educationally meaningful captions for complex classroom scenes and teaching toys.
This paper proposes and rigorously evaluates TW-DP-FedAvg, a privacy-preserving federated learning framework tailored for rural Tamil Nadu's medical insurance sector, demonstrating that while differential privacy incurs a measurable accuracy cost compared to centralized training, the approach remains statistically equivalent and viable under India's new data protection regulations.
This paper introduces a novel mothership and multi-drone routing problem for spatial marine management, proposing a metaheuristic solution that achieves near-optimal results with significantly reduced computation time compared to exact methods, thereby enabling practical-scale coordination for applications like coral reef restoration.
This paper introduces Meta-PPO, a lightweight meta-control framework that dynamically adjusts PPO hyperparameters online based on training statistics, demonstrating improved performance, efficiency, and robustness across multiple continuous-control benchmarks compared to standard PPO.
This paper presents HCLM-SmartTex, an explainable and frugal AI framework that distinguishes genuine physiological anomalies from smart-textile sensor degradation and artifacts by optimizing for human-centered decision costs rather than raw accuracy, utilizing a synthetic simulator for training and achieving high reliability with minimal computational footprint.
This paper introduces HPG-Rec, a novel, training-free framework that effectively addresses data sparsity and the semantic gap in recommender systems by combining controlled LLM candidate generation, rank-aware semantic scoring, dual popularity modeling, and MLP-based fusion to achieve significant performance improvements over state-of-the-art baselines.
This study proposes a two-layer ontology-based framework that classifies complex concrete pouring radio logs into seven semantic axes to derive hazard signals, control actions, and situation states, demonstrating high agreement rates in rule-based extraction while identifying limitations in local LLM approaches for future validation.
This paper introduces an ultra-lightweight geometric method for language identification in micro-texts that encodes character frequencies into compact Euclidean spaces, demonstrating superior robustness against data sparsity and historical variations compared to traditional baselines while offering significant computational efficiency for edge computing.
This paper demonstrates that a contradiction-aware, loop-engineered AI agent skill strictly outperforms conventional web search in generating high-stakes, decision-grade study-abroad strategies, achieving 100% validity and comprehensive output coverage compared to the baseline's 0% validity and fragmented results.
This position-style secondary analysis argues that colon CAD systems should adopt a detection-first, grading-second workflow and be evaluated using stage-specific metrics, demonstrating through reanalysis of existing data that such an approach better aligns with clinical pathology workflows and reveals that the majority of grading errors are clinically less severe adjacent-grade swaps while highlighting the critical impact of disease prevalence on positive predictive value.