Spatial

IJCAI2020 图相关论文集

◇◆丶佛笑我妖孽 提交于 2021-02-17 07:57:31
↑公众号关注 “Graph-AI” 专注于 图机器学习 IJCAI2020 图相关论文集 “ IJCAI(International Joint Conference on Artificial Intelligence,国际人工智能联合会议),是人工智能领域中最主要的学术会议之一,是CCF A类会议。 “ 本文将该会议与图相关的文章进行了整理,IJCAI接受的论文分为主赛道和特别赛道,都进行了大致的分类:计算机视觉、数据挖掘等。但因为本文是关注图方面的,所以按照图的种类来进行筛选。 “ 全部收录论文地址:http://static.ijcai.org/2020-accepted_papers.html Main track 图卷积网络 MR-GCN: Multi-Relational Graph Convolutional Networks based on Generalized Tensor Product LSGCN: Long Short-Term Traffic Prediction with Graph Convolutional Networks Multi-Class Imbalanced Graph Convolutional Network Learning Multi-View Attribute Graph Convolution Networks for

一文解读 | 计算机视觉中的注意力机制

回眸只為那壹抹淺笑 提交于 2021-02-13 02:55:46
点击上方“ 迈微电子研发社 ”,选择“ 星标★ ”公众号 重磅干货,第一时间送达 之前在看DETR这篇论文中的self_attention,然后结合之前实验室组会经常提起的注意力机制,所以本周时间对注意力机制进行了相关的梳理,以及相关的源码阅读了解其实现的机制. 一、注意力机制(attention mechanism) attention机制可以它认为是一种资源分配的机制,可以理解为对于原本平均分配的资源根据attention对象的重要程度重新分配资源,重要的单位就多分一点,不重要或者不好的单位就少分一点,在深度神经网络的结构设计中,attention所要分配的资源基本上就是权重了。 视觉注意力分为几种,核心思想是基于原有的数据找到其之间的关联性,然后突出其某些重要特征,有通道注意力,像素注意力,多阶注意力等,也有把NLP中的自注意力引入。 二、自注意力(self-attention) 参考文献: Attention is All you Need http://papers.nips.cc/paper/7181-attention-is-all-you-need 参考资料: zhuanlan.zhihu.com/p/48 GitHub: git hub.com/huggingface/ 自注意力有时候也称为内部注意力,是一个与单个序列的不同位置相关的注意力机制

32篇深度学习与遥感论文推荐

血红的双手。 提交于 2021-02-11 17:26:03
深度学习与遥感论文推荐 期刊论文推荐 1. Yuan, Q., Shen, H., Li, T., Li, Z., Li, S., Jiang, Y., … Zhang, L. ( 2020 ). Deep learning in environmental remote sensing: Achievements and challenges. Remote Sensing of Environment , 241, 111716. 2. Cunha, R. L. F. and Silva, B.: ESTIMATING CROP YIELDS WITH REMOTE SENSING AND DEEP LEARNING, ( 2020 ), ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci ., IV-3/W2-2020, 59–64. 3. Mohan, A., Singh, A. K., Kumar, B., & Dwivedi, R. ( 2020 ). Review on remote sensing methods for landslide detection using machine and deep learning. Transactions on Emerging Telecommunications

论文速览|有论文开始关注你的取消关注行为了(清华AAAI20)

做~自己de王妃 提交于 2021-02-11 13:38:28
Title: Mining Unfollow Behavior in Large-Scale Online Social Networks via Spatial-Temporal Interaction Paper: https://www.aaai.org/ojs/index.php/AAAI/article/view/5358 GitHub: https://github.com/wuhaozhe/Unfollow-Prediction 来源: AAAI 20 机构: 清华大学 在线社交网络(OSN)通过两种普遍的行为来发展:关注和取消关注,分别表示建立关系和解除关系。社会网络演化研究主要着重于关注行为,而取消关注行为在很大程度上被忽略了。挖掘取消关注行为具有挑战性,因为用户对取消关注行为的决定不仅受到用户属性(例如信息和互惠)的简单组合的影响,但也受他们之间复杂的交互作用的影响。同时,先前的数据集很少包含足够的记录来推断这种复杂的相互作用。为了解决这些问题,我们首先构建一个大规模的真实世界的微博数据集,记录了180万中国用户的详细帖子内容和关系动态。 接下来,我们将用户的属性定义为两类:空间属性(例如,用户的社会角色)和时间属性(例如,用户的帖子内容)。利用构建的数据集,我们系统地研究了用户的空间和时间属性之间的交互作用如何导致取消关注行为。之后,我们提出了UMHI

Crating Kernel density estimate for polygon in R

不想你离开。 提交于 2021-02-11 12:29:30
问题 I Have a shapefile of polygons and another one of points that are distributed over the polygons. I would like to create a kernel density estimate for each polygon based on the points it contains. unfortunately I was only able to create squared KDEs with the kde2d function from the MASS package. I would like the KDEs to be shaped as the polygons. Any suggestions? kde1 <- kde2d(poly$X, poly$Y, n=100,) enter image description here 回答1: You can use the spatstat package for this. Here is an

Analyzing spatial data between two points in R using a very large data set

北城以北 提交于 2021-02-11 05:02:17
问题 This is my first time writing code in R from scratch and I'm struggling with how to approach it. I'm looking at turtle nests and their proximity to light sources (i.e. houses, light poles, etc.) to determine how often a light source is within a given radius of a nest. These are both very large data sets (hundreds of thousands of rows) so the code will likely need to run a loop for each nest position. GPS coordinates for both data sets are in decimal degrees. The nest data is essentially

Subset spatial points with a polygon

人走茶凉 提交于 2021-02-10 12:50:13
问题 I have a SpatialPolygonsDataFrame (spolydf) and a SpatialPointsDataFrame (spointdf). The layers have different extents, but overlap. I can select points that fall within the polygon using fall.within.poly <- spointdf[spolydf,] How do I select points that fall outside the polygon? have tried fall.outside.poly <- spointdf[-spolydf,] but doesn't work. I'm mmissing something simple - any help please. 回答1: It's a bit late but I had the same issue today so I though that I would post my solution

ResNet、Faster RCNN、Mask RCNN是专利算法吗?盘点何恺明参与发明的专利

爱⌒轻易说出口 提交于 2021-02-08 13:22:39
点击上方“ 3D视觉工坊 ”,选择“星标” 干货第一时间送达 前段时间OpenCV正式将SIFT算法的实现从Non-free模块移到主库,因SIFT专利到期了(专利授权后,从申请日开始有20年的保护期)。 美国林肯总统称" 专利制度是给天才之火浇上利益之油 ",专利保护了申请人的利益,促进了科技的进步,但现在也有越来越多的滥用专利权对产业机构敲诈的案例出现。 SIFT 专利权的终结让我们不得不思考,还有哪些著名的算法被申请了专利?对于做研究的朋友来说不需要考虑这个问题,专利算法依然可以参考、复现、对比,但对于产业界朋友就不得不确认清楚:项目中有没有可能使用了别人专利保护算法。 作为计算机视觉领域当今翘楚,两度获得CVPR 最佳论文奖的何恺明大佬有很多论文都具有重大影响力。其部分论文引用数据: 残差网络 ResNet 被引用 51939 次、目标检测算法 Faster RCNN 被引用 20291 次、实例分割算法 Mask RCNN 被引用 7249 次,暗通道去雾被引用 4274 次,这些知名的算法有成百上千的开源实现,也肯定被大量的商业公司使用,有没有被申请专利? 想想 ResNet 如果被申请专利,那恐怕大多数使用深度学习的商业公司都在侵权了!价值难以估计! 带着这样的疑问,CV君检索了所有与“ Kaiming He ”相关的已经公开的授权专利和专利申请

Performance of Multiple Joins

∥☆過路亽.° 提交于 2021-02-08 11:34:33
问题 Greetings Overflowers, I need to query against objects with many/complex spacial conditions. In relational databases that is translated to many joins (possibly 10+). I'm new to this business and wondering whether to go with MS SQL Server 2008 R2 or Oracle 11g or document-based solutions such as RavenDB or simply go with some spacial database (GIS)... Any thoughts ? Regards UPDATE: Thank you all for your answers. Would anybody opt for document/spatial databases ? My database would consist of

Why does rasterToPoints generate an error on first call but not second?

筅森魡賤 提交于 2021-02-08 07:32:20
问题 I have some code that loops over a list of study IDs ( ids ) and turns them into separate polygons/spatial points. On the first execution of the loop it produces the following error: Error in (function (x) : attempt to apply non-function This is from the raster::rasterToPoints function. I've looked at the examples in the help section for this function and passing fun=NULL seems to be an acceptable method (filters out all NA values). All the values are equal to 1 anyways so I tried passing a