Global-Local Contextual Progressive Expansion Network for Martian Landslide Segmentation in Multimodal Remote Sensing Imagery
本文针对火星滑坡分割问题,提出了一种结合上下文渐进层扩展特征提取和基于Transformer上下文推理的U形网络TransCPLES,以提高分割精度和计算效率。
本文针对火星滑坡分割问题,提出了一种结合上下文渐进层扩展特征提取和基于Transformer上下文推理的U形网络TransCPLES,以提高分割精度和计算效率。
论文提出TreeRedux方法,通过在执行器上进行最终聚合操作来减少Apache Spark中树聚合时驱动器的计算和内存需求。
该研究通过结合冻结的基础模型和两次视觉-语言模型查询,提高了零样本航空图像分割的准确性,解决了基础模型直接应用时遗漏重要特征的问题。
Current safety alignment mechanisms in large language models fail to account for implicit contextual cues—such as commonsense knowledge and social norms—that are not explicitly stated, rendering them vulnerable to pragmatic attacks and creating a fundamental mismatch with human language understanding. This work introduces the novel concept of the “pragmatic attack surface,” integrating principles from pragmatics and adversarial attack techniques to devise a method that manipulates implicit context to induce harmful model outputs. The proposed approach significantly outperforms existing attack baselines across multiple mainstream open- and closed-source large language models, achieving substantially higher attack success rates. These findings expose systemic vulnerabilities in current safety mechanisms and establish a new direction for research on aligning large language models with human intent.
This study investigates financial risk spillovers in critical mineral investments and their dynamic linkages with energy markets, carbon emissions, and other macroeconomic variables. Utilizing daily data from 2013 to 2023, it pioneers the use of exchange-traded funds (ETFs) for critical minerals—rather than physical commodity prices—to construct a time-varying parameter vector autoregressive (TVP-VAR) model. Integrating dynamic connectedness and net spillover measures, the analysis uncovers time-varying interaction mechanisms among seven mineral ETFs, energy markets, carbon markets, and investor sentiment. The findings reveal that high-ESG-rated assets act predominantly as net transmitters of risk, while cobalt and aluminum ETFs serve as primary sources of shocks; conversely, WTI crude oil and carbon emission futures largely function as net receivers. The COVID-19 pandemic triggered a structural shift in these spillover roles, offering investors actionable insights for hedge strategies grounded in financial network positions.
本文针对火星滑坡分割问题,提出了一种结合上下文渐进层扩展特征提取和基于Transformer上下文推理的U形网络TransCPLES,以提高分割精度和计算效率。
论文提出TreeRedux方法,通过在执行器上进行最终聚合操作来减少Apache Spark中树聚合时驱动器的计算和内存需求。
该研究通过结合冻结的基础模型和两次视觉-语言模型查询,提高了零样本航空图像分割的准确性,解决了基础模型直接应用时遗漏重要特征的问题。
Current safety alignment mechanisms in large language models fail to account for implicit contextual cues—such as commonsense knowledge and social norms—that are not explicitly stated, rendering them vulnerable to pragmatic attacks and creating a fundamental mismatch with human language understanding. This work introduces the novel concept of the “pragmatic attack surface,” integrating principles from pragmatics and adversarial attack techniques to devise a method that manipulates implicit context to induce harmful model outputs. The proposed approach significantly outperforms existing attack baselines across multiple mainstream open- and closed-source large language models, achieving substantially higher attack success rates. These findings expose systemic vulnerabilities in current safety mechanisms and establish a new direction for research on aligning large language models with human intent.
This study investigates financial risk spillovers in critical mineral investments and their dynamic linkages with energy markets, carbon emissions, and other macroeconomic variables. Utilizing daily data from 2013 to 2023, it pioneers the use of exchange-traded funds (ETFs) for critical minerals—rather than physical commodity prices—to construct a time-varying parameter vector autoregressive (TVP-VAR) model. Integrating dynamic connectedness and net spillover measures, the analysis uncovers time-varying interaction mechanisms among seven mineral ETFs, energy markets, carbon markets, and investor sentiment. The findings reveal that high-ESG-rated assets act predominantly as net transmitters of risk, while cobalt and aluminum ETFs serve as primary sources of shocks; conversely, WTI crude oil and carbon emission futures largely function as net receivers. The COVID-19 pandemic triggered a structural shift in these spillover roles, offering investors actionable insights for hedge strategies grounded in financial network positions.