01 2020 档案

摘要:From: Stanford University; Jure Leskovec, citation 6w+; Problem: subsequence clustering. Challenging: discover patterns is challenging because it requ 阅读全文
posted @ 2020-01-31 00:41 keeps_you_warm 阅读(402) 评论(0) 推荐(0) 编辑
摘要:here. 7个问题 第一,What’s the problem? 你要做的这个问题是什么?一句话就应该回答。任何一个问题都可以在任意短的时间内,向任何人讲到你希望达到的程度。如果你没有做到,不是因为你做的问题很高精尖,是你的功夫下的不够。例如我在听一个报告,五分钟还没有听懂他在说什么,不是因为我笨 阅读全文
posted @ 2020-01-30 21:30 keeps_you_warm 阅读(293) 评论(0) 推荐(1) 编辑
摘要:from Eric Jang Normalizing flows transform simple densities (like Gaussians) into rich complex distributions that can be used for generative models, R 阅读全文
posted @ 2020-01-30 20:46 keeps_you_warm 阅读(286) 评论(0) 推荐(0) 编辑
摘要:PROBLEM: OmniAnomaly multivariate time series anomaly detection + unsupervised 主体思想: input: multivariate time series to RNN > capture the normal patte 阅读全文
posted @ 2020-01-30 00:44 keeps_you_warm 阅读(1499) 评论(0) 推荐(0) 编辑
摘要:Problem: multi-horizon probabilistic forecasting tasks; Propose an end-to-end framework for multi-horizon time series forecasting, with temporal atten 阅读全文
posted @ 2020-01-30 00:08 keeps_you_warm 阅读(920) 评论(0) 推荐(0) 编辑
摘要:from attention mechanism Attention is one component of a network’s architecture, and is in charge of managing and quantifying the interdependence. Bet 阅读全文
posted @ 2020-01-29 17:43 keeps_you_warm 阅读(219) 评论(0) 推荐(0) 编辑
摘要:KDD: Knowledge Discovery and Data Mining (KDD) Insititute: 复旦大学,中科大 Problem: time series prediction; modelling extreme events; overlook the existence 阅读全文
posted @ 2020-01-29 02:43 keeps_you_warm 阅读(489) 评论(0) 推荐(0) 编辑
摘要:1. 泛读和精度的区别和迭代: 泛读: 1个月之内,读50篇论文,进行粗读,了解多维时间序列信号,有哪些research problem, challenges, research groups, research people. 之后进行精度: 了解方法细节,如何建立的模型,模型和reality之 阅读全文
posted @ 2020-01-28 21:29 keeps_you_warm 阅读(113) 评论(0) 推荐(0) 编辑
摘要:1. how to measure the uncertainty of prediction model or data analysis? 2. 阅读全文
posted @ 2020-01-28 16:39 keeps_you_warm 阅读(94) 评论(0) 推荐(0) 编辑
摘要:From Tutte institute for mathematics and computing Problem: dimension reduction Theoretical foundations: At a high level, UMAP uses local manifold app 阅读全文
posted @ 2020-01-27 23:13 keeps_you_warm 阅读(970) 评论(0) 推荐(0) 编辑
摘要:materials: 1. Dimension Reduction - IsoMap 阅读全文
posted @ 2020-01-27 22:27 keeps_you_warm 阅读(145) 评论(0) 推荐(0) 编辑
摘要:The 10th international conference on machine vision; C类 Methodology: 非主流方法 2 stages: 1. convert time series data to recurrence plot. 数值*时间长度 > 时间长度*时间 阅读全文
posted @ 2020-01-27 19:09 keeps_you_warm 阅读(185) 评论(0) 推荐(0) 编辑
摘要:Dimensionality in statistics refers to how many attributes a dataset has. For example, healthcare data is notorious for having vast amounts of variabl 阅读全文
posted @ 2020-01-24 19:05 keeps_you_warm 阅读(232) 评论(0) 推荐(0) 编辑
摘要:from Django official document; Django 最初被设计用于具有快速开发需求的新闻类站点,目的是要实现简单快捷的网站开发。 from 编写你的第一个 Django 应用,第 1 部分. the first Django startProject 1. create ou 阅读全文
posted @ 2020-01-23 18:50 keeps_you_warm 阅读(364) 评论(0) 推荐(0) 编辑
摘要:pip install altair pip install altair_viewer Altair is a declarative statistical visualization library for Python, based on Vega and Vega-Lite, and th 阅读全文
posted @ 2020-01-23 17:24 keeps_you_warm 阅读(195) 评论(0) 推荐(0) 编辑
摘要:tasks: 1. unsupervised knn https://scikit-learn.org/stable/modules/neighbors.html#unsupervised-neighbors 2. isomap 3. largevis 4. line 5.umap 1. how t 阅读全文
posted @ 2020-01-22 17:59 keeps_you_warm 阅读(1158) 评论(0) 推荐(0) 编辑
摘要:MDS, multidimensional scaling, 线性降维方法, 目的就是使得降维之后的点两两之间的距离尽量不变(也就是和在原是空间中对应的两个点之间的距离要差不多)。只是 MDS 是针对欧氏空间设计的,对于距离的计算也是使用欧氏距离来完成的。如果数据分布在一个流形上的话,欧氏距离就不适 阅读全文
posted @ 2020-01-22 00:16 keeps_you_warm 阅读(222) 评论(0) 推荐(0) 编辑
摘要:1. The difference between classification and clustering. from here. Classification: supervised learning with labels. Clustering: unsupervised learning 阅读全文
posted @ 2020-01-21 23:40 keeps_you_warm 阅读(2569) 评论(0) 推荐(0) 编辑
摘要:website; A visual survey of visualization techniques for time-oriented data. 1. Left pannel filter of visualization techniques; Number of variables: u 阅读全文
posted @ 2020-01-21 22:41 keeps_you_warm 阅读(190) 评论(0) 推荐(0) 编辑
摘要:1. 你一定要看计算机领域的文章. 如果你是看一堆应用,你最终还是会不知道怎么做. 从计算机到energy是降维打击, 当你学习了计算机的hot skill,再去做应用很容易. 2. 搞研究的思路: 大量阅读文章,然后形成思路,然后justify每一步的合理性和可行性. 你需要critical th 阅读全文
posted @ 2020-01-21 20:31 keeps_you_warm 阅读(212) 评论(0) 推荐(0) 编辑
摘要:Don't look back. Don't hesitate, just do it. t-SNE原理 from here. 1. tsne is strictly used for visualization. and we can only see things in up to 3 dime 阅读全文
posted @ 2020-01-20 22:34 keeps_you_warm 阅读(184) 评论(0) 推荐(0) 编辑
摘要:1. Run part of my Python file. here Ctrl+Alt+E 2. pycharm cell mode 3. 阅读全文
posted @ 2020-01-20 20:14 keeps_you_warm 阅读(153) 评论(0) 推荐(0) 编辑
摘要:Python-louvain Package pip install python-louvain #Drawing partition Method 1: Method 2: Supplementary knowledge: 1. what is the partition of graphs. 阅读全文
posted @ 2020-01-20 20:10 keeps_you_warm 阅读(1061) 评论(0) 推荐(0) 编辑
摘要:1. 要严格遵循自己的作息时间,不能周六日放松,但是一到周六日就像变了个人是的,以往的严律己不现. 有没有什么办法,让自己周末也能严以律己, 而非侥幸觉得自己辛苦五天了,就应该休息. from 我是如何克服惰性的. 一、人人都有惰性 如果有人告诉你他完全克服了自己的惰性,他肯定也是吹牛逼的。 记住: 阅读全文
posted @ 2020-01-20 17:30 keeps_you_warm 阅读(204) 评论(0) 推荐(0) 编辑
摘要:1. from here. diagonalReturn specified diagonals. diagflatCreate a 2-D array with the flattened input as a diagonal. traceSum along diagonals. triuUpp 阅读全文
posted @ 2020-01-19 00:27 keeps_you_warm 阅读(598) 评论(0) 推荐(0) 编辑
摘要:HeatManDataLake 1. schema: hfors tables 2. schema: bronderslev not useful. the data is not about the energy consumption. Public Data. Public data webs 阅读全文
posted @ 2020-01-16 23:59 keeps_you_warm 阅读(128) 评论(0) 推荐(0) 编辑
摘要:1. docker command 阅读全文
posted @ 2020-01-16 18:35 keeps_you_warm 阅读(75) 评论(0) 推荐(0) 编辑
摘要:here 1. find files in Linux Use find from the command line to locate a specific file by name or extension. The following example searches for *.err fi 阅读全文
posted @ 2020-01-16 17:20 keeps_you_warm 阅读(98) 评论(0) 推荐(0) 编辑
摘要:Official 教程 关闭postgresql服务 PostgreSQL帐号 1. PostgreSQL 用户帐号和操作系统用户帐号是不同的,系统用户帐号是postgres. Use the operating system user postgres to create your databas 阅读全文
posted @ 2020-01-15 21:46 keeps_you_warm 阅读(342) 评论(2) 推荐(0) 编辑
摘要:1. There's two types of python environment in pycharm: virtualenv Environment conda environment For conda environment, you can access them by conda co 阅读全文
posted @ 2020-01-15 20:07 keeps_you_warm 阅读(156) 评论(0) 推荐(0) 编辑
摘要:增加新的视图: 1. 在app/view.py下增加调用视图函数 def horizonG(request): return render(request, 'horizonG.html') 2. 在/template下添加horizonG.html文件 3. demo/urls.py下添加url调 阅读全文
posted @ 2020-01-15 19:13 keeps_you_warm 阅读(128) 评论(0) 推荐(0) 编辑
摘要:Problem: 1. We can't find patterns in full attribute space, and patterns may only be found in smaller subspaces. 2. Pattern Trails is an interactive v 阅读全文
posted @ 2020-01-15 00:27 keeps_you_warm 阅读(111) 评论(0) 推荐(0) 编辑
摘要:Problem: get an overall picture of how ego-networks evolve is a common challenging task. Existing techniques: inspect the evolution patterns of ego-ne 阅读全文
posted @ 2020-01-14 04:35 keeps_you_warm 阅读(153) 评论(0) 推荐(0) 编辑
摘要:However: The conventional visual mapping maps each data attribute onto a single visual channel Purpose: investigate composite visual mapping综合可视映射: ma 阅读全文
posted @ 2020-01-13 23:57 keeps_you_warm 阅读(137) 评论(0) 推荐(0) 编辑
摘要:观念: 1. 把可视化的东西拾起来, 毕竟是做那个出身的. 2. 可视化里面时序数据,时空数据一直都是非常重要的. 3. know your data永远是最重要的一步, 我想更好的方式是,数据驱动,而不是方法驱动,因为方法在好可能不是和我们的问题. ?? 数据驱动 VS 方法驱动. 4. 时序序列 阅读全文
posted @ 2020-01-13 20:12 keeps_you_warm 阅读(123) 评论(0) 推荐(0) 编辑
摘要:Purpose Implement a good user aggregation and classification. or to assess the interrelation patterns between user profiles. Data i. daily temperature 阅读全文
posted @ 2020-01-13 20:06 keeps_you_warm 阅读(93) 评论(0) 推荐(0) 编辑
摘要:1. 共轭复数 在数学中,复数的共轭复数(常简称共轭)是对虚部变号的运算,因此一个复数 的复共轭是 将复数理解为复平面,则复共轭无非是对实轴的反射。复数的复共轭有时也表为 2. 矩阵A的复数共轭A*定义为[A*]ij = aij* 3. 矩阵的运算 矩阵的乘法不满足交换律。 4. 阅读全文
posted @ 2020-01-12 01:14 keeps_you_warm 阅读(368) 评论(0) 推荐(0) 编辑
摘要:Purpose detect the dynamics in time series of their correlation Methodology 1. calculate correlation coefficients 2. Network construction: i. node: AA 阅读全文
posted @ 2020-01-11 00:26 keeps_you_warm 阅读(144) 评论(0) 推荐(0) 编辑
摘要:1. 矩阵分解可以用来解决什么方法, 以及how? 利用矩阵分解来解决实际问题的分析方法很多,如PCA(主成分分析)、ICA(独立成分分析)、SVD(奇异值分解)、VQ(矢量量化)等。在所有这些方法中,原始的大矩阵V被近似分解为低秩的V=WH形式。这些方法的共同特点是,因子W和H中的元素可为正或负, 阅读全文
posted @ 2020-01-10 21:48 keeps_you_warm 阅读(465) 评论(0) 推荐(0) 编辑
摘要:1. Python Web(1):建立第一个Web项目 直接将d3 index.html代码拷贝进index.html中即可. 2. D3入门教程; D3.js的v5版本入门教程 阅读全文
posted @ 2020-01-09 23:47 keeps_you_warm 阅读(165) 评论(0) 推荐(0) 编辑
摘要:Improvement can be done in fulture:1. the algorithm of constructing network from distance matrix. 2. evolution of sliding time window3. the later proc 阅读全文
posted @ 2020-01-09 12:06 keeps_you_warm 阅读(231) 评论(0) 推荐(0) 编辑
摘要:在第22行中,由于tseires 变成了嵌入元组, 所以max(tseries[ta + 1:tb]) 会取下标最大的值, 而非最大的第二个元素. [(0, 0.19024852355156963), (1, 0.6660417262541884), (2, 0.395523497583831), 阅读全文
posted @ 2020-01-08 16:41 keeps_you_warm 阅读(176) 评论(0) 推荐(0) 编辑
摘要:重点: 1.The constructed graph inherits several properties of the series in its structure. periodic series --> regular graphs这句话是错的,periodic series 转化成的图 阅读全文
posted @ 2020-01-07 17:48 keeps_you_warm 阅读(241) 评论(0) 推荐(0) 编辑

点击右上角即可分享
微信分享提示