8123 papers
The paper proposes Graph-of-Mark (GoM), a novel pixel-level visual prompting technique that overlays scene graphs onto images to capture object relationships, thereby significantly enhancing the spatial reasoning and zero-shot performance of multimodal language models.
This paper introduces a system-level inference optimization for Diffusion Transformer-based video generation that employs a sequence-parallel Causal-RoPE mechanism and operator fusion to overcome memory and latency bottlenecks, achieving near real-time speeds and sub-second first-frame latency on an eight-GPU cluster.
This paper reveals that Chain-of-Thought prompting often underperforms direct answering in medical visual question answering due to a "medical perception bottleneck," and proposes training-free grounding interventions to restore visual accuracy and improve model reasoning.
This paper introduces SIL-GPO, a reinforcement learning framework that combines graph attention networks with self-imitation learning to optimize the joint deployment and routing of heterogeneous edge AI and microservices, significantly reducing end-to-end latency and improving resource utilization compared to existing methods.
The paper proposes CalibFusion, a Transformer-based differentiable calibration framework that learns to implicitly refine Radar-Camera extrinsics end-to-end to overcome the challenges of textureless, cluttered water-surface environments and significantly improve fusion-based 2D object detection.
This paper introduces ERP-RiskBench, a leakage-safe ensemble learning framework for detecting financial risks in ERP systems that combines hybrid risk definitions, rigorous time- and group-aware validation to prevent data leakage, and a stacking gradient boosting model to deliver reproducible, interpretable, and operationally grounded fraud detection results.
This paper investigates whether semantic noise initialization, known to improve image diffusion models, transfers to text-to-video generation, finding that while it shows a slight positive trend on temporal metrics, it does not significantly outperform standard Gaussian noise due to weak or unstable signals in the noise space.
AutoFigure-Edit is an end-to-end system that generates fully editable, high-quality scientific illustrations from long-form text with flexible style adaptation via reference images, leveraging long-context understanding and native SVG support to overcome limitations in editability and efficiency found in existing automated tools.
This paper proposes a hybrid few-shot learning model integrating Siamese and Prototypical Networks with Grad-CAM-based Explainable AI to achieve high-accuracy, interpretable classification of maize, rice, and wheat leaf diseases under limited data conditions.
This paper addresses the limitations of current Large Vision-Language Models in deep chart research by proposing Parallel Relative Policy Optimization (PRPO) to resolve training conflicts and constructing the MCDR-Bench evaluation framework to enable objective assessment of complex reasoning capabilities.
The paper introduces MultiGen, a novel diffusion-based game engine that incorporates an explicit, persistent external memory to enable user-editable world structures and support coherent, real-time multiplayer interactions, overcoming the limitations of conventional next-frame prediction models.
This paper introduces VB, a novel benchmark designed to evaluate vision-language models' ability to determine image visibility and appropriately abstain from answering when evidence is insufficient, utilizing controlled minimal edits and specialized metrics to reveal that top-tier models like GPT-4o and Gemini 3.1 Pro significantly outperform open-source alternatives in confidence-aware accuracy and perspective reasoning.
The paper introduces "Narrative Weaver," a novel framework that achieves controllable, long-range visual consistency in generative AI by integrating multimodal narrative planning with a dynamic memory bank, validated through extensive experiments and a newly released e-commerce advertising dataset.
This paper introduces a method that enhances medical Vision-Language Models by using sequential eye-tracking data as supervision to train dedicated gaze tokens, enabling the models to mimic radiologists' visual search patterns and achieve state-of-the-art performance in both in-domain and out-of-domain medical reasoning tasks.
This paper proposes a novel approach to mining data-aware temporal specifications from execution traces by combining Syntax Guided Synthesis with a finite-prefix interpretation of Temporal Stream Logic (TSL), enabling the robust and sample-efficient synthesis of reactive programs that capture both data transformations and temporal behaviors.
The paper introduces ATLAS, a reinforcement finetuning framework that enables small language models to effectively navigate large toolspaces by learning adaptive context acquisition and execution strategies, thereby achieving frontier-level performance with significantly reduced parameter and context budgets.
This paper proposes a hierarchical planning approach that integrates supplemental geostationary satellite data to extend lookahead horizons for Dynamic Targeting missions, demonstrating up to a 41% performance improvement over traditional onboard-only planners, particularly in scenarios with sparsely distributed targets.
The paper introduces ProtAlign, a contrastive learning framework that unifies protein sequence and structure representations into a shared embedding space, thereby enabling cross-modal retrieval and improving downstream tasks like function annotation and stability estimation.
This paper introduces Universal Watermark Presence Detection (UWPD), a novel task for identifying invisible watermarks without prior algorithm knowledge, supported by the UniFreq-100K dataset and the Frequency Shield Network (FSNet) model that achieves superior zero-shot detection by dynamically amplifying high-frequency watermark signals while suppressing semantic content.
This paper proposes a bi-directional feedback fusion framework that integrates human activity embeddings with dual-timescale temporal modules to significantly improve the accuracy and interpretability of indoor CO2 and PM2.5 forecasting compared to traditional data-driven models.