4157 papers
This paper proposes a cost-effective collaborative framework that combines a BERT-based proxy active learning strategy with full-parameter Flan-T5 training to accurately detect Self-Admitted Technical Debt using only 30% of the labeled data required by traditional full-supervision methods.
This study establishes a demographic benchmark for pediatric forearm fracture localization using the PediURF dataset, demonstrating that age and sex alone provide moderate predictive information (macro AUC ~0.70) comparable to simple logistic regression, while more complex machine learning algorithms offered no significant improvement and class imbalance handling primarily influenced decision thresholds rather than the intrinsic predictive power of the metadata.
This study proposes a scalable, data-driven framework utilizing LightGBM regression on the DEAM dataset to accurately predict music-induced emotional valence and arousal from acoustic features, achieving strong performance metrics that support applications in personalized recommendations and music therapy.
This paper introduces Boundary-Stabilized SMOTE (BS-SMOTE), a domain-constrained oversampling technique that integrates feature-specific interpolation and physiological feasibility clipping to significantly improve the accuracy and interpretability of rare adverse event predictions in data-scarce, imbalanced clinical settings, as demonstrated by its superior performance in predicting peri-anesthetic hypoxemia in veterinary anesthesia.
This paper introduces AICOF, an AI-driven framework that integrates MLP traffic estimation, GNN topology learning, and PPO-based reinforcement learning to optimize adaptive routing across diverse processor interconnection networks, significantly enhancing throughput, latency, energy efficiency, and fault resilience compared to conventional methods.
This study demonstrates that artificial neural networks trained on the micro-tonal pitch structure of Iranian classical music spontaneously develop hidden unit representations dominated by Fourier phase spaces, confirming that this frequency range serves as a universal candidate code for musical pitch structure beyond Western tonal systems.
This paper argues that archival organizations should overcome their hesitation to share collections for AI training by reframing the loss of item-level provenance as a competitive advantage and adopting strategic policies—such as tailored agreements, RAG technology, and advocacy for public-good contributions—to mitigate risks while fostering collaboration.
The paper introduces SAMVT-Net, a compact supervised model that leverages source-aware multi-view token learning and explicit lead identity to effectively handle heterogeneous ECG data for both heartbeat and record-level classification tasks.
This paper demonstrates that the validity and stability of Shapley explanations for tabular foundation models are critically dependent on the choice of removal operator relative to the intended estimand and are significantly influenced by in-context examples, necessitating explicit specification of these factors for reliable interpretation.
This paper demonstrates that while graph neural networks initially appear to outperform sequential baselines in detecting AI-generated text, rigorous controlled auditing reveals their superior performance is an artifact of untuned regularization and that their graph structure merely re-encodes information already captured by tuned embedding probes, offering no genuine architectural advantage.