基于机器学习的稀疏样本下的土壤有机质估算方法

采用GRNN（Generalized Regression Neural Network）和RF（Random Forest）2种机器学习方法构建土壤有机质预测模型,以提高稀疏样本情况下的土壤有机质估算精度。依据北京市大兴区农用地2007年的土壤有机质采样数据,按MMSD准则（Minimization of the Mean of the Shortest Distances）抽稀为8种不同采样密度的样本（分别为2703、1352、676、339、169、85、43、22个样本）,分别采用GRNN、RF和Ordinary kriging对各采样密度下的未知采样点进行预测,采用交叉检验的方式验证各采样密度下未知样点的预测精度。随着采样点密度的下降,样点间的空间自相关性逐渐减弱,半变异函数的拟和精度变差,预测点结果误差增大,预测的置信度降低。当抽稀到43个和22个采样点时,样点间的空间自相关性接近歼灭,半变异函数的决定系数较低且残差较大。普通克里格受到采样点数量和采样密度、样点的空间结构的影响比较明显,其预测精度随采样点数量的下降而下降。在85个采样点及以下时,其预测值与观测值之间没有显著的相关性。GRNN和RF的预测精度受采样密度的影响不大,其预测精度在一个较小的范围内波动,其预测值围绕观测值在一定阈值空间内震荡波动,具有较好的相关性,在85个及以下的采样密度时,预测精度相对普通克里格有较大的提升。普通克里格法不适合在稀疏样本条件下空间插值计算,尤其是在空间自相关性比较弱的情况下。机器学习模型能充分学习土壤间环境信息、样点空间邻近效应信息,兼顾属性相似性和空间自相关,具有更好的稳定性和适应性,不容易受到采样点数量、构型和采样密度等因素的影响,即使在采样点空间自相关性很弱的情况下也能做出稳定预测精度。     

基于Item2Vec负采样优化的专题地图产品个性化推荐方法研究

建立适用于专题地图产品检索的用户偏好推荐模型是提高专题地图质量的有效方式之一,在专题地图产品推荐场景中,存在严重的内容冷启动和评论数据稀疏问题,现有的推荐算法无法为特定类用户推荐不同特征的专题地图产品,导致用户从专题地图中获取偏好信息受到限制。因此,本文构建基于负采样的连续词袋模型和基于Word2Vec的Item2Vec相结合的用户偏好推荐方法,用于专题地图产品推荐。① 计算用户行为日志文件中交互行为数据的隐性评分,以代替专题地图推荐信息流场景中稀疏的用户评论数据;② 基于负采样的连续词袋模型提取目标专题地图的前后地图序列感知特征信息,通过控制正负样本比例为1:2,提升目标专题地图潜在评分的预测精度;③ 通过Item2Vec将带有用户行为特征信息的专题地图映射到向量空间,计算用户对专题地图的相似度矩阵,根据用户偏好程度完成推荐。在构建的专题地图评分实验数据集Thematic CMaps和4个公开验证数据集MovieLens上的测试结果表明:与LFM、Personal Rank、Content Based和SVD 4种传统推荐算法相比,本文所提方法可有效提高潜在评分的预测精度,推荐性能最高达到27.85%;与以霍夫曼采样方式的Item2Vec基础方法和YouTubeNet 2种神经网络推荐算法相比,评分预测精度有一定提高,且推荐性能不断提升,最高达到2.97%和5.78%。以经典算法奇异值分解（SVD）为例,将MovieLens-20M数据集切分后,在数据量不断增大的数据子集中,本文所用方法的评分预测精度和性能均优于SVD方法。     

浅谈如何提高测绘工程专业学生对遥感原理与方法课程学习的兴趣

结合笔者在中南大学为期9年的遥感原理与方法课程教学实践,分析了中南大学测绘工程专业学生在该课程教学方面的学习态度和课程认识方面的现状,提出在教学过程初期加强学生对遥感历史文化及名人传记的学习,在课程教学期间自己查询专业应用前景及专业最新动态新闻,在上课期间主动向老师提问3种方法,以考核方式计入学生平时成绩,以期增强学生对该课程的学习兴趣和主动学习能力及独立思考能力,为后续遥感专业相关课程的学习奠定一定的基础。在近几年的教学实践过程中,发现该方法有效地提高了学生对该专业课的学习兴趣,学习效果超越往届学生,实践结果充分证明了该方法的有效性。     

案例推理的制图综合应用背景与方法

从挖掘和利用经验数据的角度出发,提出了基于案例推理的智能化制图综合方法。首先探讨了基于算法和模型的自动综合实现方式所面临的瓶颈,并提出了采用制图综合案例推理辅助制图综合的思想;然后设计了制图综合专家案例的三元表示方法,以及获取、使用和管理流程;最后对实现该基于案例推理的智能化制图综合方法的五个关键性步骤进行了分析,并深入阐述该方法的原理和可行性,提出了基于案例推理的智能综合辅助系统架构。     

A Comparative Study of Various Supervised Learning Approaches to Selective Omission in a Road Network

Selective omission is necessary for road network generalisation. This study investigates the use of supervised learning approaches for selective omission in a road network. To be specific, at first, the properties to measure the importance of a road in the network are viewed as input attributes, and the decision of such a road is retained or not at a specific scale is viewed as an output class; then, a number of samples with known input and output are used to train a classifier; finally, this classifier can be used to determine whether other roads to be retained or not. In this study, a total of nine supervised learning approaches, i.e., ID3, C4·5, CRT, Random Tree, support vector machine (SVM), naive Bayes (NB), K-nearest neighbour (KNN), multilayer perception (MP) and binary logistic regression (BLR), are applied to three road networks for selective omission. The performances of these approaches are evaluated by both quantitative assessment and visual inception. Results show that: (1) in most cases, these approaches are effective and their classification accuracy is between 70% and 90%; (2) most of these approaches have similar performances, and they do not have any statistically significant difference; (3) but sometimes, ID3 and BLR performs significantly better than NB and SVM; NB and KNN perform significantly worse than MP, SVM and BLR.     

高分辨率遥感影像的深度学习变化检测方法

为提升高分辨率遥感影像的变化检测精度,提出一种利用深度学习的变化检测方法。在预处理的基础上,利用顾及邻域信息的改进变化矢量分析算法和灰度共生矩阵算法获取影像间光谱和纹理变化,并通过设置自适应采样区间提取最可能的变化和未变化区域样本。构建并训练包含标签层的高斯伯努利深度限制玻尔兹曼机模型,以提取变化和未变化区域深层特征,从而有效辨别变化区域。通过WorldView-3与Pléiades-1影像的试验表明本文方法在变化检测精度方面优于对比方法。     

面向高光谱图像分类的半监督空谱判别分析

为充分利用高光谱图像蕴藏的空间信息提升分类精度,提出了面向高光谱图像分类的半监督空谱判别分析(S3 DA)算法。考虑高光谱图像数据集的空间一致性,首先利用少量标记样本定义类内散度矩阵,保存数据集同类像元的光谱近邻结构;再利用无标记样本定义空间近邻像元散度矩阵,揭示像元间的空间近邻结构和地物的空间分布结构信息。S3 DA既保持数据集在光谱域的可分性,又保存了无标记样本蕴藏的空间域近邻结构,增强了同类像元和空间近邻像元在投影子空间的聚集性,从而提升分类性能。在PaviaU和Indian Pines数据集的试验表明,总体分类精度分别达到81.50%和71.77%。与传统的光谱方法比较,该算法能有效提升高光谱图像数据集的地物分类精度。     

递进式项目教学法在“GIS程序设计”课程中的探索与实践

"GIS程序设计"作为地理信息科学专业的必修课,对学生专业能力的培养起到重要作用,同时GIS开发也是目前地理信息科学专业就业的主要方向。文中结合工程实践,通过校园地理信息系统、森林火灾扩散模拟分析与决策系统、小组答辩自选专题地理信息系统3个从简单到复杂的GIS项目,运用递进式的项目教学方法和团队协作、课上实践、课上提问、课内外答疑、课后作业、编写项目文档等措施对该课程进行改革探索和实践,并取得较好的教学效果。     

Should Geography Educators Adopt Electronic Textbooks?

Significant changes in textbook publishing are at college instructors' door. Textbook representatives are directed to mesmerize potential clients with eye-catching displays and statements about digital literacy of the next generation of college students. Based on a one-year pilot study, in 2014–2015, of an electronic textbook, this commentary shares real experiences with the digital tools, contextualizes the shift in textbook publishing and research on student outcomes, and provides insight in navigating through the Brave New World of digital texts and adaptive learning tools.     

In search of meaningful assessment in the university curriculum: the case for culturally relevant pedagogy

ABSTRACT

The poor performance of Pacific students at university is a concern for every level of society. Conventional models of teaching, learning and assessment have overlooked the cultural background of students, yet the effects of this oversight have been disastrous, alienating and disempowering. Studies of and developments in education in the Pacific and elsewhere offer opportunities to rethink the methods of teaching, learning and assessment of Pacific students in ways that are congruent with their home culture. Informed by the principle of so’a lau pule (the Samoan concept of consensus through consultation and conversation) and ethnographic research methods, this study describes an attempt to rethink ways of teaching, learning and assessing student performance in a third-year course on Resource Conservation and Management in the School of Geography, Earth Science and Environment at the University of the South Pacific. Introduced in 2007, the Student Innovative Contribution to Knowledge (SICK) allows students to participate in key decisions concerning teaching, learning and assessment. SICK is grounded in the principles of inclusiveness and participation. It takes account of the skills and abilities of Pacific students and is aligned with the philosophical basis of Pacific cultures. One hundred and fifty-eight students took part in this study over a 5-year period. Analysis was based on students’ written reviews of the semester’s work, in-depth conversation with individual students and the end-of-the-course evaluation. The findings show that, while students embrace the need for non-conventional forms of teaching, learning and assessment, the most frequently cited responses for non-participation are poor time management and varying perceptions of what is considered an innovative assessment. This study shows that understanding the cultural background of students is critical to creating culturally inclusive learning environments.