Clueing up LLMs with Tool-Augmented Deductive Reasoning
研究通过在经典桌游Clue的多代理版本中引入工具增强的方法,使用可能性矩阵来提高大型语言模型在多步骤推理任务中的逻辑一致性和表现。
研究通过在经典桌游Clue的多代理版本中引入工具增强的方法,使用可能性矩阵来提高大型语言模型在多步骤推理任务中的逻辑一致性和表现。
本文提出ProgResViT,通过逐步提高图像分辨率和模型宽度来适应性地处理图像分类问题,以实现更好的准确率与计算量之间的平衡。
This study addresses the challenge of distinguishing inundated land from permanent water bodies in Sentinel-1 SAR imagery by systematically evaluating the performance of convolutional neural networks (U-Net, U-Net++, DeepLabV3) and vision Transformers (SegFormer-b0/b1/b2) on multi-class flood segmentation tasks. To assess spatial generalization, the authors introduce a scene-partitioning evaluation strategy and employ interpretability techniques—including Grad-CAM and uncertainty estimation—to elucidate differences in model decision mechanisms. Experimental results demonstrate that SegFormer-b2 significantly outperforms CNN-based models on the ETCI dataset; although this advantage narrows after fine-tuning on Sen1Floods11, SegFormer-b2 maintains superior performance in spatially fragmented flood scenarios. Interpretability analyses further confirm that its activations are more tightly aligned with genuine flood signatures, underscoring the potential and reliability of Transformer architectures for SAR-based flood mapping.
Existing approaches struggle to effectively integrate the complementary information from spaceborne optical and SAR imagery, limiting the accuracy of flood mapping. This work proposes a cross-modal multi-task distillation network that simultaneously achieves high-fidelity optical image reconstruction and flood region segmentation within a unified latent space. The method innovatively combines a lightweight optical teacher guidance module, multi-scale windowed cross-attention, FiLM-based conditional modulation, and gated residual connections, alongside a dual-decoder architecture and a hydrologically aware loss function incorporating Charbonnier SSIM, edge-preserving FFT magnitude, and Dice-BCE alignment. Evaluated on SEN1FLOODS11, DEEPFLOOD, and SEN12MS datasets, the approach significantly outperforms current fusion methods in terms of PSNR, SSIM, and LPIPS, markedly enhancing flood mapping quality.
This study addresses structural and conceptual gaps in AI safety governance—particularly regulatory fragmentation, insufficient global coordination, and limited participation from the Global South—that hinder cross-jurisdictional interoperability. Through a comparative analysis of China, South Korea, Singapore, and the United Kingdom across three high-risk domains—autonomous vehicles, education, and cross-border data flows—the research systematically integrates ethical, legal, and technical governance dimensions. It identifies convergences and divergences across seven key aspects: governance objectives, regulatory authorities, ethical principles, enforcement mechanisms, domain-specific frameworks, technical standards, and critical risks. Aligning interoperability pathways with the UN-endorsed Global Digital Compact and relevant UN resolutions, the study offers actionable policy recommendations that balance local contexts with global cooperation, thereby advancing an inclusive, effective, and trustworthy AI safety governance framework.
研究通过在经典桌游Clue的多代理版本中引入工具增强的方法,使用可能性矩阵来提高大型语言模型在多步骤推理任务中的逻辑一致性和表现。
本文提出ProgResViT,通过逐步提高图像分辨率和模型宽度来适应性地处理图像分类问题,以实现更好的准确率与计算量之间的平衡。
This study addresses the challenge of distinguishing inundated land from permanent water bodies in Sentinel-1 SAR imagery by systematically evaluating the performance of convolutional neural networks (U-Net, U-Net++, DeepLabV3) and vision Transformers (SegFormer-b0/b1/b2) on multi-class flood segmentation tasks. To assess spatial generalization, the authors introduce a scene-partitioning evaluation strategy and employ interpretability techniques—including Grad-CAM and uncertainty estimation—to elucidate differences in model decision mechanisms. Experimental results demonstrate that SegFormer-b2 significantly outperforms CNN-based models on the ETCI dataset; although this advantage narrows after fine-tuning on Sen1Floods11, SegFormer-b2 maintains superior performance in spatially fragmented flood scenarios. Interpretability analyses further confirm that its activations are more tightly aligned with genuine flood signatures, underscoring the potential and reliability of Transformer architectures for SAR-based flood mapping.
Existing approaches struggle to effectively integrate the complementary information from spaceborne optical and SAR imagery, limiting the accuracy of flood mapping. This work proposes a cross-modal multi-task distillation network that simultaneously achieves high-fidelity optical image reconstruction and flood region segmentation within a unified latent space. The method innovatively combines a lightweight optical teacher guidance module, multi-scale windowed cross-attention, FiLM-based conditional modulation, and gated residual connections, alongside a dual-decoder architecture and a hydrologically aware loss function incorporating Charbonnier SSIM, edge-preserving FFT magnitude, and Dice-BCE alignment. Evaluated on SEN1FLOODS11, DEEPFLOOD, and SEN12MS datasets, the approach significantly outperforms current fusion methods in terms of PSNR, SSIM, and LPIPS, markedly enhancing flood mapping quality.
This study addresses structural and conceptual gaps in AI safety governance—particularly regulatory fragmentation, insufficient global coordination, and limited participation from the Global South—that hinder cross-jurisdictional interoperability. Through a comparative analysis of China, South Korea, Singapore, and the United Kingdom across three high-risk domains—autonomous vehicles, education, and cross-border data flows—the research systematically integrates ethical, legal, and technical governance dimensions. It identifies convergences and divergences across seven key aspects: governance objectives, regulatory authorities, ethical principles, enforcement mechanisms, domain-specific frameworks, technical standards, and critical risks. Aligning interoperability pathways with the UN-endorsed Global Digital Compact and relevant UN resolutions, the study offers actionable policy recommendations that balance local contexts with global cooperation, thereby advancing an inclusive, effective, and trustworthy AI safety governance framework.