基于A- CGAN的深反射地震数据随机噪声压制方法研究

基于深度学习的地震数据噪声压制方法是当前地震数据去噪处理的重要方向。深度学习方法突破了传统滤波处理的局限,在对常规地震数据的噪声压制中表现出效率高、信噪分离效果好的特点。但针对深部弱有效反射数据,当前的深度学习方法特征提取能力有限,难以取得较好的去噪效果。笔者等结合深反射地震数据特点,针对当前深度学习噪声压制方法在特征提取及对数据集依赖上的局限,提出了基于注意力循环生成对抗网络(Attention Cycle- Consistent Generative Adversarial Networks,A- CGAN)的深反射地震数据随机噪声压制方法。借助循环一致生成对抗网络(Cycle- Consistent Generative Adversarial Networks,Cycle- GAN)的域映射思想,降低对数据集的要求。为了构建适用于深反射地震数据的去噪网络,从3个方面对Cycle- GAN进行改进：在Cycle- GAN的生成器（去噪器）中加入残差结构和注意力机制,用于加深网络深度和提高其特征提取能力;在Cycle- GAN的鉴别器中使用块判决,提高鉴别精度和准确度;在损失函数部分加入感知一致性损失函数,提升网络模型恢复纹理细节信息的能力。通过合成地震数据和实际深反射地震数据测试,验证了优化算法的有效性,体现了良好的应用价值。     

预测有效波高的深度学习模型研究

研究基于RNN、LSTM、GRU深度学习模型,针对NOAA浮标数据集中的44013、44014、44017浮标的数据,通过斯皮尔曼相关性分析提高模型预测效果。实验结果表明,在进行相关性分析后,S-RNN、S-LSTM、 S-GRU的预测效果均比原始RNN、LSTM、GRU模型预测效果好。此外,提出一种基于LSTM的LSTM-Attention 波高预测模型,并进行相关实验,量化LSTM-Attention模型的预测效果,实验结果表明LSTM-Attention模型有更好的预测效果。为评估模型的泛化能力,研究还提出了一种采用邻近浮标数据进行学习,预测浮标缺失数据的方 法。实验结果表明,该方法的预测精度可以达到97.93%。本研究为海浪预测提供了新的方法和思路,也为未来深 度学习模型在海浪预测中的应用提供了参考。     

可解释性长短期记忆模型用于预测湖泊总磷浓度变化

对湖泊总磷的变化预测和来源识别对水资源调度和流域生态治理有着重要的意义,然而复杂的生化反应和水动力条件导致的非平稳性给湖泊总磷浓度的准确预测带来极大的困难。为克服这一挑战,本文引入了基于加权回归的季节趋势分解(seasonal and trend decomposition using Loess,STL)技术和夏普利加法(SHapley additive exPlanations,SHAP)结合长短期记忆网络(long short-term memory neural network,LSTM)和门控循环单元(gated recurrent unit,GRU)构建了一个可解释的预测框架,以增强对湖泊总磷浓度演变的预测并提高其可解释性。研究表明:(1)在骆马湖总磷浓度的预测中,该框架拥有较好的预报精度(R²=0.878),优于LSTM和卷积长短期记忆模型(convolutional neural networks and long short term memory network,CNN-LSTM)。当预测时间步长增加到8 h时,该框架有效提高了总磷浓度的预测精度,平均相对误差和均方根误差分别降低了47.1%和33.3%。从预测趋势来看,骆马湖在汛期的总磷平均浓度为0.158 mg/L,相较于非汛期的平均浓度,增加了202.1%。(2)运河来水是骆马湖总磷浓度最重要的影响因素,贡献权重为60.0%,并且不同断面(三湾、三场)的污染源受水动力、气象等因素的影响存在显著的时空差异。本文凸显了神经网络模型在预警水体污染方面的可实施性,并且为提高传统神经网络的学习能力和可解释性的开发与验证提供了重要方向。     

基于Faster R-CNN的食品图像检索和分类

面向食品领域的图像检索和分类等方面的研究成为多媒体分析和应用领域越来越受关注的研究课题之一.当前的主要研究方法基于全图提取视觉特征,但由于食品图像背景噪音的存在使得提取的视觉特征不够鲁棒,进而影响食品图像检索和分类的性能.为此,本文提出了一种基于Faster R-CNN网络的食品图像检索和分类方法.首先通过Faster R-CNN检测图像中的候选食品区域,然后通过卷积神经网络（CNN）方法提取候选区域的视觉特征,避免了噪音的干扰使得提取的视觉特征更具有判别力.此外,选取来自视觉基因库中标注好的食品图像集微调Faster R-CNN网络,以保证Faster R-CNN食品区域检测的准确度.在包括233类菜品和49 168张食品图像的Dish-233数据集上进行实验.全面的实验评估表明：基于Faster R-CNN食品区域检测的视觉特征提取方法可以有效地提高食品图像检索和分类的性能.     

从地理信息系统到虚拟地理环境的认知转变

自20世纪60年代初以来,地理信息系统引领了地理信息的分析与服务。然而,由于地理学家对于地理过程开展多维模拟的需求,同时许多大型工程需要对于不同方案的后果事先开展模拟评估,为此中国学者们在世纪相交之际开始了对于虚拟地理环境的探索。本文从地理信息系统到虚拟地理环境的认识上的转变着手,分析了虚拟地理环境的演进过程、当前的阶段定位与挑战,着重探讨了虚拟地理环境研究的新态势：基于虚拟地理环境的地理空间认知研究、虚拟地理环境与实验地理学新方向、大数据背景下的虚拟地理认知实验方法和虚拟地理环境与地理知识工程,并给出最新研究思路。     

基于约束Markov决策过程的初至自动识别技术

随着地震数据采集技术的进步,地震数据量日益增加,全自动、高精度的地震初至走时拾取技术受到了更加广泛的关注.本文将初至拾取看作特征空间内带约束的Markov决策过程,在奖励函数空间,按一定准则全局寻优获得积累奖励值最大的路径,从而达到在高维空间自动拾取初至信息的目的.同时,状态值函数中包含与距离相关的折扣因子γ,使Markov决策过程拾取初至能够考虑地震数据的横向连续性,并且回避地震数据中的坏道信息.在此基础上,本文方法进一步引入受空间几何信息约束的动作（Actions）和转移概率（Transitions Probability）,从而降低了对起始状态和折扣因子选取的难度,让地震数据初至走时拾取更加准确和自动化.实际数据测试结果表明,在初至能量较弱（信噪比较低）情况或浅层存在相邻较近复杂波形时,本文提出的约束Markov算法仍能准确地进行初至走时的自动拾取,并且具有一定的质量监控能力,让拾取结果更有物理意义.

     

Site characterization model using least‐square support vector machine and relevance vector machine based on corrected SPT data (Nc)

Statistical learning algorithms provide a viable framework for geotechnical engineering modeling. This paper describes two statistical learning algorithms applied for site characterization modeling based on standard penetration test (SPT) data. More than 2700 field SPT values (N) have been collected from 766 boreholes spread over an area of 220 sqkm area in Bangalore. To get N corrected value (N_c), N values have been corrected (N_c) for different parameters such as overburden stress, size of borehole, type of sampler, length of connecting rod, etc. In three‐dimensional site characterization model, the function N_c=N_c (X, Y, Z), where X, Y and Z are the coordinates of a point corresponding to N_c value, is to be approximated in which N_c value at any half‐space point in Bangalore can be determined. The first algorithm uses least‐square support vector machine (LSSVM), which is related to a ridge regression type of support vector machine. The second algorithm uses relevance vector machine (RVM), which combines the strengths of kernel‐based methods and Bayesian theory to establish the relationships between a set of input vectors and a desired output. The paper also presents the comparative study between the developed LSSVM and RVM model for site characterization. Copyright © 2009 John Wiley & Sons, Ltd.     

迁移学习支持下的高分影像山地滑坡灾害解译模型

无人机低空遥感是近年来新兴的一种快速获取灾情信息的手段,如何利用无人机高分影像构建滑坡灾害解译模型是实现快速自动解译滑坡的关键。针对该问题,对比了多种影像特征提取方法,将迁移学习(TL)特征和支持向量机(SVM)引入到构建滑坡灾害自动解译模型中,提出了一种TL支持下的高分影像滑坡灾害解译模型。选取5·12汶川地震及4·20芦山地震系列无人机影像构建了滑坡灾害样本库并进行了实验,TL特征方法整体分类准确度ACC为95%,ROC达到0.98,识别准确率达到97%。结果表明,所提方法可用于高分影像滑坡自动解译,同时可用于大面积高分影像中快速山地滑坡灾害定位及检测。     

Historical paths to current unrest: Extending the temporal lens in analysing geographies of agrarian change and conflict

The proliferation of soybean monoculture and its expansion into previously marginal territories in Argentina have resulted in land conflicts and attacks on peasant communities that possess land without having secured tenure. However, the rapid and intensive modernisation and industrialisation of agriculture production has had differentiated effect on different territories in Argentina. This article posit that understanding these differences requires an extended temporal analysis. While social research tends to focus on the present or recent past, this paper joins calls for greater attention to historical context and processes when examining current social dynamics and geographies. Focusing on the province of Santiago del Estero in Northwest Argentina, this paper provides a particular temporal-thematic approach for conducting such a historical path analysis, which here focuses on the province’s changing agrarian structures and political shifts that resulted in a complex, irregular and subsequently unstable land tenure system.     

Industrial and geographical mobility of workers during industry decline: The Swedish and German shipbuilding industries 1970–2000

This article follows the industry employment histories of all individuals who at some point have been affiliated with the declining German or the dismantling Swedish shipbuilding industry during 1970–2000. We analyse the situation of the individual workers leaving shipbuilding, investigating the extent to which they were employed at all, tended to move to related sectors within or outside the region, and whether such moves were beneficial for the individuals. Combining insights from labour geography and redundancy studies with evolutionary economic geography, we find remarkably similar results for the West German and Swedish cases. Our findings indicate a notable impact of the regional industry structure on the labour market outcomes for workers leaving shipbuilding. This suggests that more attention should be devoted to the specific structures of the absorptive capacity of regional labour markets. The findings are discussed within the context of a mature industry.