2787 papers
This paper introduces PriseBot, an AI-powered conversational agent leveraging NLP and RAG technologies to guide users through the PRISE process for developing iStar extensions, demonstrating its effectiveness and usability through the creation of the iStar4Bot extension and validation studies involving both novice and experienced researchers.
This paper presents a training-free, self-calibrating theory for blind deconvolution that derives a closed-form estimator for blur-kernel scale by leveraging extreme value statistics of natural images and the additive property of kernel radii, achieving superior accuracy and downstream deblurring performance compared to existing heuristics and supervised baselines.
This paper introduces Q-PDPL, a hybrid quantum-classical optimization framework that integrates VQE, QAOA, and Quantum Graph Attention Networks within a Dantzig-Wolfe decomposition scheme to efficiently solve large-scale, stochastic Pickup and Delivery Problems with Lockers, demonstrating superior cost reduction and scalability compared to classical algorithms.
This paper proposes the Semantic-Retrieval-Reasoning Pipeline (SRRP), a hybrid LLM framework that enhances grocery bundle recommendations by detecting multi-intent user needs to generate diverse, explainable, and optimally sized product bundles with human-aligned naming and reasoning.
This paper proposes the Decoupled Diffusion-GAN Restoration framework (D2R), a two-stage method that combines structural reconstruction and texture refinement to effectively restore ancient Chinese mountain-pattern bronze mirrors by enforcing their rigid geometric regularities, thereby outperforming generic inpainting models and providing a non-invasive tool for archaeological analysis.
This paper introduces MPC-SNN, a novel convolutional spiking neural network that achieves state-of-the-art accuracy on the DVS128 Gesture benchmark using strictly local, backpropagation-free learning by enabling neurons to predict the feedforward inputs of their peers through a learned lateral weight matrix.
This paper proposes a Federated Model Optimization (FMO) framework that integrates gradient compression, reliability-weighted aggregation, and hardware-aware neural architectures to enable privacy-preserving, low-latency real-time anomaly detection for smart drug delivery edge devices, thereby overcoming the limitations of conventional cloud-centric approaches.
The paper proposes UniPAS, a strategy-level reinforcement learning framework that eliminates the classifier bottleneck in database synchronization by embedding urgency awareness directly into the reward function, enabling a deep Q-network to dynamically navigate the Pareto frontier between fairness and urgency without relying on coarse event categorization.
This paper introduces the ZERO framework, an operational governance model that shifts AI data-center sustainability decision-making upstream by enforcing multi-resource admissibility constraints before workload execution, thereby prioritizing holistic resource availability over post-hoc efficiency metrics.
This paper introduces RAM-GenAI, a practical rapid assessment method designed to evaluate responsible generative AI use in scientific research within low- and middle-income contexts like Uganda by focusing on five key domains: transparency, verification, data responsibility, human oversight, and reproducibility.