The urban transit fare structure and level can largely affect passengers’travel behavior and route choices.The commonly used transit fare policies in the present transit network would lead to the unbalanced transit a...The urban transit fare structure and level can largely affect passengers’travel behavior and route choices.The commonly used transit fare policies in the present transit network would lead to the unbalanced transit assignment and improper transit resources distribution.In order to distribute transit passenger flow evenly and efficiently,this paper introduces a new distance-based fare pattern with Euclidean distance.A bi-level programming model is developed for determining the optimal distance-based fare pattern,with the path-based stochastic transit assignment(STA)problem with elastic demand being proposed at the lower level.The upper-level intends to address a principal-agent game between transport authorities and transit enterprises pursing maximization of social welfare and financial interest,respectively.A genetic algorithm(GA)is implemented to solve the bi-level model,which is verified by a numerical example to illustrate that the proposed nonlinear distance-based fare pattern presents a better financial performance and distribution effect than other fare structures.展开更多
Public authorities frequently mandate public or private agencies to manage their renewable natural resources.Contrary to the agency,which is an expert in renewable natural resource management,public authorities usuall...Public authorities frequently mandate public or private agencies to manage their renewable natural resources.Contrary to the agency,which is an expert in renewable natural resource management,public authorities usually ignore the sustainable level of harvest.In this note,we first model the contractual relationship between a principal,who owns the renewable natural resource,and an agent,who holds private information on its sustainable level of harvest.We then look for the Pareto-optimal allocations.In the situation of an imperfect information setting,we find that the Pareto-optimal contracting depends on the probability that the harvesting level stands outside the sustainability interval.The information rent held by the agent turns out to be unavoidable,such that stepping outside the sustainability interval implies the possibility of depletion of the renewable natural resource.This,in turn,compromises the maintenance of the ecological balance in natural ecosystems.展开更多
[目的]现有的语义变化检测方法对于遥感影像的局部和全局特征利用不足,且忽略了不同时相间的时空依赖性,导致土地覆盖语义分类结果不精确。此外,检测的变化对象存在边缘模糊问题,检测结果和实际变化区域的一致性有待提升。[方法]针对这...[目的]现有的语义变化检测方法对于遥感影像的局部和全局特征利用不足,且忽略了不同时相间的时空依赖性,导致土地覆盖语义分类结果不精确。此外,检测的变化对象存在边缘模糊问题,检测结果和实际变化区域的一致性有待提升。[方法]针对这些挑战,受具有长序列处理能力的视觉状态空间模型(Vision State Space Model, VSSM)启发,本文提出了一种融合卷积神经网络(Convolutional Neural Networks, CNN)与VSSM的语义变化检测网络CVS-Net。该网络有效结合了CNN的局部特征提取优势与VSSM捕捉长距离依赖关系的能力,并嵌入基于VSSM的双向时空关系建模模块以引导网络深入理解影像间的时空变化逻辑关系。此外,为增强模型对变化对象边缘的识别精度,提出了边缘感知强化分支,通过联合拉普拉斯算法和多任务架构自动集成边界信息,增强模型对于变化地物的形状模式的学习能力。[结果]在SECOND和FZSCD数据集上,将本方法与HRSCD.str4、Bi-SRNet、ChangeMamba、ScanNet及TED五种主流的SCD方法进行对比。实验结果显示,本方法的语义变化检测性能优于其他对比方法,验证了本方法的有效性。在SECOND数据集上实现了23.95%的分离卡帕系数(Sek)和72.89%的平均交并比(mIoU),在FZ-SCD数据集上的SeK达到23.02%, mIoU达到72.60%。消融实验结果中,随着在基础网络中添加各模块,SeK值逐步提升至21.26%、23.04%和23.95%,证明了CVS-Net中各模块的有效性。[结论]本方法可有效提升了语义变化检测的属性和几何精度,可为城市可持续发展、土地资源管理等应用领域提供技术参考和数据支撑。展开更多
针对现有端到端自动驾驶模型输入数据类型单一导致预测精确度低的问题,选取RGB图像、深度图像和车辆历史连续运动状态序列作为多模态输入,并利用语义信息构建一种基于时空卷积的多模态多任务(Multimodal Multitask of Spatial-temporal ...针对现有端到端自动驾驶模型输入数据类型单一导致预测精确度低的问题,选取RGB图像、深度图像和车辆历史连续运动状态序列作为多模态输入,并利用语义信息构建一种基于时空卷积的多模态多任务(Multimodal Multitask of Spatial-temporal Convolution,MM-STConv)端到端自动驾驶行为决策模型,得到速度和转向多任务预测参量。首先,通过不同复杂度的卷积神经网络提取场景空间位置特征,构建空间特征提取子网络,准确解析场景目标空间特征及语义信息;其次,通过长短期记忆网络(LSTM)编码-解码结构捕捉场景时间上、下文特征,构建时间特征提取子网络,理解并记忆场景时间序列信息;最后,采用硬参数共享方式构建多任务预测子网络,输出速度和转向角的预测值,实现对车辆的行为预测。基于AirSim自动驾驶仿真平台采集虚拟场景数据,以98200帧虚拟图像及对应的车辆速度和转向角标签作为训练集,历经10000次训练周期、6h训练时长后,利用真实驾驶场景数据集BDD100K进行模型的测试与验证工作。研究结果表明:MMSTConv模型的训练误差为0.1305,预测精确度达到83.6%,在多种真实驾驶场景中预测效果较好;与现有其他主流模型相比,该模型综合场景空间信息与时间序列信息,在预测车辆速度和转向角方面具有明显的优势,可提升模型的预测精度、稳定性和泛化能力。展开更多
目标表观建模是基于稀疏表示的跟踪方法的研究重点,针对这一问题,提出一种基于判别性局部联合稀疏表示的目标表观模型,并在粒子滤波框架下提出一种基于该模型的多任务跟踪方法 (Discriminative local joint sparse appearance model bas...目标表观建模是基于稀疏表示的跟踪方法的研究重点,针对这一问题,提出一种基于判别性局部联合稀疏表示的目标表观模型,并在粒子滤波框架下提出一种基于该模型的多任务跟踪方法 (Discriminative local joint sparse appearance model based multitask tracking method,DLJSM).该模型为目标区域内的局部图像分别构建具有判别性的字典,从而将判别信息引入到局部稀疏模型中,并对所有局部图像进行联合稀疏编码以增强结构性.在跟踪过程中,首先对目标表观建立上述模型;其次根据目标表观变化的连续性对采样粒子进行初始筛选以提高算法的效率;然后求解剩余候选目标状态的联合稀疏编码,并定义相似性函数衡量候选状态与目标模型之间的相似性;最后根据最大后验概率估计目标当前的状态.此外,为了避免模型频繁更新而引入累积误差,本文采用每5帧判断一次的方法,并在更新时保留首帧信息以减少模型漂移.实验测试结果表明DLJSM方法在目标表观发生巨大变化的情况下仍然能够稳定准确地跟踪目标,与当前最流行的13种跟踪方法的对比结果验证了DLJSM方法的高效性.展开更多
基金the Humanities and Social Science Foundation of the Ministry of Education of China(Grant No.20YJCZH121).
文摘The urban transit fare structure and level can largely affect passengers’travel behavior and route choices.The commonly used transit fare policies in the present transit network would lead to the unbalanced transit assignment and improper transit resources distribution.In order to distribute transit passenger flow evenly and efficiently,this paper introduces a new distance-based fare pattern with Euclidean distance.A bi-level programming model is developed for determining the optimal distance-based fare pattern,with the path-based stochastic transit assignment(STA)problem with elastic demand being proposed at the lower level.The upper-level intends to address a principal-agent game between transport authorities and transit enterprises pursing maximization of social welfare and financial interest,respectively.A genetic algorithm(GA)is implemented to solve the bi-level model,which is verified by a numerical example to illustrate that the proposed nonlinear distance-based fare pattern presents a better financial performance and distribution effect than other fare structures.
基金financially supported by a grant overseen by the French National Forestry Office through the Forests for Tomorrow International Teaching and Research Chair(Convention particulière n°1/2013)supported by the French National Research Agency through the Laboratory of Excellence ARBRE,a part of the Investments for the Future Program(ANR 11--LABX-0002-01).
文摘Public authorities frequently mandate public or private agencies to manage their renewable natural resources.Contrary to the agency,which is an expert in renewable natural resource management,public authorities usually ignore the sustainable level of harvest.In this note,we first model the contractual relationship between a principal,who owns the renewable natural resource,and an agent,who holds private information on its sustainable level of harvest.We then look for the Pareto-optimal allocations.In the situation of an imperfect information setting,we find that the Pareto-optimal contracting depends on the probability that the harvesting level stands outside the sustainability interval.The information rent held by the agent turns out to be unavoidable,such that stepping outside the sustainability interval implies the possibility of depletion of the renewable natural resource.This,in turn,compromises the maintenance of the ecological balance in natural ecosystems.
文摘[目的]现有的语义变化检测方法对于遥感影像的局部和全局特征利用不足,且忽略了不同时相间的时空依赖性,导致土地覆盖语义分类结果不精确。此外,检测的变化对象存在边缘模糊问题,检测结果和实际变化区域的一致性有待提升。[方法]针对这些挑战,受具有长序列处理能力的视觉状态空间模型(Vision State Space Model, VSSM)启发,本文提出了一种融合卷积神经网络(Convolutional Neural Networks, CNN)与VSSM的语义变化检测网络CVS-Net。该网络有效结合了CNN的局部特征提取优势与VSSM捕捉长距离依赖关系的能力,并嵌入基于VSSM的双向时空关系建模模块以引导网络深入理解影像间的时空变化逻辑关系。此外,为增强模型对变化对象边缘的识别精度,提出了边缘感知强化分支,通过联合拉普拉斯算法和多任务架构自动集成边界信息,增强模型对于变化地物的形状模式的学习能力。[结果]在SECOND和FZSCD数据集上,将本方法与HRSCD.str4、Bi-SRNet、ChangeMamba、ScanNet及TED五种主流的SCD方法进行对比。实验结果显示,本方法的语义变化检测性能优于其他对比方法,验证了本方法的有效性。在SECOND数据集上实现了23.95%的分离卡帕系数(Sek)和72.89%的平均交并比(mIoU),在FZ-SCD数据集上的SeK达到23.02%, mIoU达到72.60%。消融实验结果中,随着在基础网络中添加各模块,SeK值逐步提升至21.26%、23.04%和23.95%,证明了CVS-Net中各模块的有效性。[结论]本方法可有效提升了语义变化检测的属性和几何精度,可为城市可持续发展、土地资源管理等应用领域提供技术参考和数据支撑。
文摘针对现有端到端自动驾驶模型输入数据类型单一导致预测精确度低的问题,选取RGB图像、深度图像和车辆历史连续运动状态序列作为多模态输入,并利用语义信息构建一种基于时空卷积的多模态多任务(Multimodal Multitask of Spatial-temporal Convolution,MM-STConv)端到端自动驾驶行为决策模型,得到速度和转向多任务预测参量。首先,通过不同复杂度的卷积神经网络提取场景空间位置特征,构建空间特征提取子网络,准确解析场景目标空间特征及语义信息;其次,通过长短期记忆网络(LSTM)编码-解码结构捕捉场景时间上、下文特征,构建时间特征提取子网络,理解并记忆场景时间序列信息;最后,采用硬参数共享方式构建多任务预测子网络,输出速度和转向角的预测值,实现对车辆的行为预测。基于AirSim自动驾驶仿真平台采集虚拟场景数据,以98200帧虚拟图像及对应的车辆速度和转向角标签作为训练集,历经10000次训练周期、6h训练时长后,利用真实驾驶场景数据集BDD100K进行模型的测试与验证工作。研究结果表明:MMSTConv模型的训练误差为0.1305,预测精确度达到83.6%,在多种真实驾驶场景中预测效果较好;与现有其他主流模型相比,该模型综合场景空间信息与时间序列信息,在预测车辆速度和转向角方面具有明显的优势,可提升模型的预测精度、稳定性和泛化能力。
文摘目标表观建模是基于稀疏表示的跟踪方法的研究重点,针对这一问题,提出一种基于判别性局部联合稀疏表示的目标表观模型,并在粒子滤波框架下提出一种基于该模型的多任务跟踪方法 (Discriminative local joint sparse appearance model based multitask tracking method,DLJSM).该模型为目标区域内的局部图像分别构建具有判别性的字典,从而将判别信息引入到局部稀疏模型中,并对所有局部图像进行联合稀疏编码以增强结构性.在跟踪过程中,首先对目标表观建立上述模型;其次根据目标表观变化的连续性对采样粒子进行初始筛选以提高算法的效率;然后求解剩余候选目标状态的联合稀疏编码,并定义相似性函数衡量候选状态与目标模型之间的相似性;最后根据最大后验概率估计目标当前的状态.此外,为了避免模型频繁更新而引入累积误差,本文采用每5帧判断一次的方法,并在更新时保留首帧信息以减少模型漂移.实验测试结果表明DLJSM方法在目标表观发生巨大变化的情况下仍然能够稳定准确地跟踪目标,与当前最流行的13种跟踪方法的对比结果验证了DLJSM方法的高效性.
