4157 papers
This paper proposes a SWOT-driven, five-aspect responsible adoption framework specifically designed to guide higher education institutions in evaluating their readiness, governance, and compliance for implementing GenAI-based cybersecurity solutions, addressing the limitations of existing standalone frameworks.
This paper proposes a lightweight, anomaly-based intrusion detection framework for IoT networks that combines a sparse autoencoder for feature extraction with a novel Dispersion-Based Majority Class Reduction (DMCR) technique to efficiently address class imbalance and hardware constraints, achieving significant reductions in training time and model size while maintaining detection quality compared to synthetic oversampling methods.
This paper proposes a reinforcement learning-based dynamic event-triggered interception strategy for unknown maneuvering targets under incomplete and asymmetric information that adaptively switches between optimal and game-based policies to reduce conservatism while ensuring system stability and excluding Zeno behavior.
This paper demonstrates that standard aggregate exact-match evaluation of SNOMED CT concept normalization obscures a significant performance gap where systems fail more often on finer-grained (NARROW) mentions than on exact-granularity ones, arguing for granularity-stratified error reporting and hierarchy-aware scoring metrics.
This paper presents a structured qualitative review of failure recovery in tool-using LLM agents by organizing a taxonomy of detection, correction, and state consistency mechanisms, analyzing nine representative works to highlight that while explicit error handling is well-explored, silent failures, unreliable self-correction, and insufficient attention to state consistency remain critical open challenges.
This paper presents a unified runtime and fidelity analysis of the Enhanced Adaptive Discriminant Compression framework, demonstrating its statistical superiority over baselines and its capability to achieve near-real-time, high-accuracy compression for both image and text data in mobile cloud computing environments.
This study demonstrates that retaining adherence-process variables which overlap with outcome definitions causes severe feature leakage, artificially inflating machine-learning model performance for predicting antiretroviral non-adherence in children with HIV, and advocates for tiered sensitivity analyses to distinguish true predictive signal from data artifacts.
This paper proposes a theoretical framework for bounded explanation deferral in intelligent systems that distinguishes between critical and deferrable information to create an auditable specification for managing explanation debt, ensuring human oversight and contestability are preserved without assuming that reduced workload inherently improves safety.
This paper critically evaluates public resume–job matching benchmarks by demonstrating that current methods often rely on data leakage and noisy labels, are vulnerable to superficial formatting artifacts like markdown, and offer only modest gains over simple keyword baselines (with LLM judges not significantly beating embeddings), thereby recommending stricter evaluation protocols including unseen-resume splits and data cleaning.
This paper introduces an auditable, anytime-valid statistical framework with executable certificates to rigorously decide whether quantized large language models are non-inferior to reference configurations across specific deployment environments, thereby controlling false-certification risks while supporting flexible, configuration-specific release assurance.