898 papers
This paper proposes an improved APF-RRT algorithm that integrates power-law decay repulsive gains, a three-level escape mechanism, and adaptive strategies to significantly enhance path planning efficiency, smoothness, and obstacle avoidance for basket-carrying robotic arms in complex agricultural environments.
This paper demonstrates that training a Small Language Model (Gemma 2B) with RLVR and GRPO to enforce calibrated abstention significantly improves clinical safety by doubling the deferral rate for ambiguous cases, despite a trade-off in raw benchmark accuracy.
This paper systematically demonstrates that in chest X-ray AI, correct predictions are frequently accompanied by inaccurate saliency map explanations, revealing a critical decoupling between prediction accuracy and localization fidelity that necessitates mandatory validation thresholds before clinical deployment.
This paper proposes a multi-agent LLM framework featuring distinct planner, executor, and verifier roles operating on a dependency-aware task graph with calibrated parameters and shared-memory mechanisms, which significantly outperforms single-agent baselines in complex task solving by improving completion rates, accuracy, and factual consistency while reducing tool errors.
The paper introduces Stratum-eval, a normative-first evaluation framework for clinical machine learning that mandates a validated Normative Specification Document defining use cases, fairness trade-offs, and stakeholder boundaries before computing any performance metrics, thereby ensuring ethical and regulatory alignment is established prior to model deployment.
This paper introduces ISVEX, a comprehensive vulnerability scoring method featuring both "Strong" and "Lite" variants designed to address the specific limitations of existing standards by quantifying unique AI risks such as model manipulation, data poisoning, and agentic vulnerabilities through a tailored multi-metric framework.
This paper proposes a parallel thinking-trace-guided auto-learning framework for autonomous robots that integrates a main thinking system with parallel Auto Learning and Obvious Learning subsystems to efficiently convert reasoning traces into executable models, thereby significantly reducing latency and high-level reasoning invocations while maintaining high task success rates.
The paper proposes SWST-Net, a semantic-aware wavelet-driven network that utilizes discrete wavelet transforms to separate and interactively reconstruct image structures and textures guided by semantic priors, achieving superior performance in image inpainting across multiple datasets.
This paper proposes and validates a decentralized, peer-to-peer entropy generator that utilizes a commit-reveal protocol and hybrid aggregation to provide statistically uniform, low-cost, and attack-resistant randomness for blockchain-based games, outperforming centralized oracles like Chainlink VRF in both economic efficiency and security.
This paper proposes the Spatio-Temporal Collaboration Network (STC-Net), which integrates frame and event modalities through an Adaptive Spiking Neuron module and a High-order Spatio-Temporal Fusion module to achieve state-of-the-art object tracking performance in challenging conditions.