833 papers
This paper proposes a novel Beam Search-Driven Dual-Path Feature Selection with Weighted Merging (BSDPFS-WM) model that integrates advanced preprocessing, multi-strategy feature selection, and stacked generalization to achieve robust, generalizable, and explainable multi-cancer detection across diverse clinical datasets.
This paper presents a comprehensive benchmark of nine encoder-decoder architectures for automated fracture radiograph captioning on a clinically curated dataset, demonstrating that the vision-language pre-trained BLIP-Base model outperforms other combinations of visual encoders and GPT-2 decoders while providing critical insights into failure modes and architectural selection for low-resource medical imaging tasks.
This paper demonstrates that while the ForesightLM-v2 model successfully learns a stable sentence-level future prediction signal, this representation fails to directly control generation behavior, revealing that its apparent benefits stem primarily from semantic reranking rather than the learned future term itself and highlighting a critical gap between learned representations and calibrated behavioral outcomes in small language models.
This paper proposes a behavior-centric framework that reconstructs Sysmon events into GUID-correlated sequences to improve MITRE ATT&CK technique classification, demonstrating that preserving process-level context outperforms temporal grouping and that a baseline SecureBERT model achieves superior results without explicit class imbalance mitigation.
This paper introduces Elemental Variational Expanse (EVE), a neuron-level variational unit that replaces deterministic hidden activations with input-dependent sampled latent states to improve probabilistic prediction and provide measurable diagnostics while maintaining compatibility with standard neural architectures.
This paper proposes GeoDualNet, the first end-to-end neural codec that jointly learns disparity vector prediction and depth map coding through a unified geometric latent space featuring five novel components, achieving significant bitrate savings and improved synthesized view quality over traditional 3D-HEVC and existing learned multiview codecs.
This paper introduces Modality-Conditional Conformal Prediction (MC²), a method that restores valid, informative coverage guarantees for multimodal medical diagnosis models under missing modality inputs by stratifying calibration according to modality-availability patterns without requiring model retraining.
This paper presents the Enhanced Passive Reader-Writer (EPRW) lock, a userspace adaptation of the original PRW lock that eliminates the incompatible heartbeat requirement for poll-mode datapaths while preserving scalability and adding critical features like lock upgrade, downgrade, and non-blocking acquisition.
This paper proposes ICATS, an Issue-Centric Autonomous Thinking Scheduling framework that enables robots in dynamic environments to prioritize internal cognitive processing by treating diverse factors like unfinished tasks and emerging risks as dynamic issues, thereby outperforming traditional fixed-baseline methods in weighted reward and status detection.
This paper presents an interactive decision support dashboard for the UAC Care System, developed using Python, Streamlit, SQLite, and Power BI, which integrates real-time monitoring, predictive analytics, and scenario planning to enhance operational visibility and evidence-based decision-making for public-sector care managers.