70 Matching Annotations
  1. Jun 2020
    1. The double backquote is used to enter in verbatim mode, which can be used as the escaping character. There are some restrictions about the * and `` syntax. They cannot not be nested, content may not start or end with whitespace: * text* is wrong, it must be separated from surrounding text by non-word characters like a space.

      Reason for following piece of comments are prefixed with 'r', being a raw-string.

          r"""Validate behavior with network_metadata.
      
          Ensure we handle ``NUMACell``\ s that have ``network_metadata`` set
          along with those where this is unset.
          ""
      
  2. Jan 2020
    1. In a small/simple scenario, the rules do not have much of an impact as all the services can communicate with each other on a single message bus and in a single cell database. However, as the deployment grows, scaling and security concerns may drive separation and isolation of the services.

      it means the cellv2 is configurable? Need steps for this.

  3. Mar 2019
    1. alembic: 不抄我怎么解释 通常我们会将我们的代码放入到某个VCS(版本控制系统)中,进行可追溯的版本管理。一个项目除了代码,通常还会有一个数据库,这个数据库可能会随着项目的演进发生变化,甚至需要可以回滚到过去的某个状态,于是一些工具将数据库的版本化也纳入了管理。 Alembic 是 Sqlalchemy 的作者实现的一个数据库版本化管理工具,它可以对基于Sqlalchemy的Model与数据库之间的历史关系进行版本化的维护。 随着软件的开发,功能的增加,方向的调整,数据库的结构也会跟着变化,那么数据库结构怎么管理呢?手动管理,实在不是程序员的办法,alembic就是为了自动化处理数据库结构的工具。会随着程序中对数据库对象的定义,半自动的修改你的数据库结构。

      简洁 明了

    1. versioned objects

      Feature:(oslo.versionedobjects): The oslo.versionedobjects library provides a generic versioned object model that is RPC-friendly, with inbuilt serialization, field typing, and remotable method calls. It can be used to define a data model within a project independent of external APIs or database schema for the purposes of providing upgrade compatibility across distributed services.

    1. Update a wrapper function to look like the wrapped function. The optional arguments are tuples to specify which attributes of the original function are assigned directly to the matching attributes on the wrapper function and which attributes of the wrapper function are updated with the corresponding attributes from the original function. The default values for these arguments are the module level constants WRAPPER_ASSIGNMENTS (which assigns to the wrapper function’s __name__, __module__ and __doc__, the documentation string) and WRAPPER_UPDATES (which updates the wrapper function’s __dict__, i.e. the instance dictionary). The main intended use for this function is in decorator functions which wrap the decorated function and return the wrapper. If the wrapper function is not updated, the metadata of the returned function will reflect the wrapper definition rather than the original function definition, which is typically less than helpful.

      name, module and doc

    1. Decorators provide a simple syntax for calling higher-order functions. By definition, a decorator is a function that takes another function and extends the behavior of the latter function without explicitly modifying it.

      Definition:

      1. calling higher-order functions
      2. take another function and extends the behavior of latter function
      3. return a function has the same prototype.
    1. python装饰器的wraps作用

      decorator is a little tricky for the argument, sometimes it the argument is the function name, some times is the argument passed in decorator function. But generally it have rule: -. If @wrap has no () in the line, the the parameter for wrap is the function, otherwise it is the argument. -. A decorator requests to return a function has the same declaration to target function.

    1. Report CPU features to the placement service

      3 drawbacks. Two for the inconsistent of using for cpu features crossing different virtual machines. and one for inefficiency of query of CPU features, reason seems that CPU features are kept in compute node, and have to access each compute node. But what's the improvements in placement? These CPU trait will be kept near the requester or be more centered, which means lesser requests in number?

    1. os-traits

      os-traits: Is that mean OpenStack is using a separate library to manage 'trait's, for the term that introduced in placement proposal?

      If we first have the concept of placement that we have the 'os-traits' library?

    1. JSON Schema is a vocabulary that allows you to annotate and validate JSON documents.

      'JSON Schema' definition: vocabulary allows you to annotate and validate JSON docs.

    1. JSON基于两种结构:json简单说就是javascript中的对象和数组,所以这两种结构就是对象和数组两种结构,通过这两种结构可以表示各种复杂的结构

      Json数据格式:数组和对象以及数值

    1. The Cambodian government has rejected a demand from a major opposition party to release its leader from detention. Prime Minister Hun Sen said Monday that Kem Sokha, leader of Cambodia's National Rescue Party, will remain detained for now

      Remain detained / detention

  4. www.thefreedictionary.com www.thefreedictionary.com
  5. www.thefreedictionary.com www.thefreedictionary.com
  6. Jun 2018
    1. 当运行我们的测试套件时,可能会发生测试结果不确定,因为测试套件依赖于系统中特定的文件可访问或具有某些特定的值。

      When running our test suite it may happen that the test result is nondeterministic because of the test suite relying on a particular file in the system being accessible or having some specific value.

    2. ./run在顶层还有一个脚本,可以使运行尚未安装的程序更容易,并且可以在gdb或Valgrind下封装各种测试的调用。

      There is also a ./run script at the top level, to make it easier to run programs that have not yet been installed, as well as to wrap invocations of various tests under gdb or Valgrind.

    3. 这些测试默认从tarball编译或配置选项--enable-expensive-tests时运行; 您也可以通过在创建时将VIR_TEST_EXPENSIVE设置为0或1来强制执行这些测试的一次性切换

      These tests default to being run when building from a tarball or with the configure option --enable-expensive-tests; you can also force a one-time toggle of these tests by setting VIR_TEST_EXPENSIVE to 0 or 1 at make time, as in:

    4. 根据开发环境中的缺省情况,某些测试会根据所需的时间与增量构建期间这些测试可能导致问题的可能性相比较而被忽略。

      Some tests are skipped by default in a development environment, based on the time they take in comparison to the likelihood that those tests will turn up problems during incremental builds.

  7. Apr 2018
    1. In Linux 4.12 kernel and newer, Intel RDT/MBA is enabled by kernel config CONFIG_INTEL_RDT.

      CMT is enabled in 4.10 and newer. MBM is enabled in 4.12 and newer.

  8. Feb 2018
    1. So far, there are three known variants of the issue: Variant 1: bounds check bypass (CVE-2017-5753) Variant 2: branch target injection (CVE-2017-5715) Variant 3: rogue data cache load (CVE-2017-5754)

      three variants

    1. national mechanism for the prevention of torture in Congress, which had [...] already been preliminarily approved by the Senate.

      preliminary ... preliminarily

    1. MADV_DONTNEED Do not expect access in the near future. (For the time being, the application is finished with the given range, so the kernel can free resources associated with it.)

      In near future...

    1. 缺页中断分为两类,一种是内存缺页中断,这种的代表是malloc,利用malloc分配的内存只有在程序访问到得时候,内存才会分配;另外就是硬盘缺页中断,这种中断的代表就是mmap,利用mmap映射后的只是逻辑地址,当我们的程序访问时,内核会将硬盘中的文件内容读进物理内存页中,这里我们就会明白为什么mmap之后,访问内存中的数据延时会陡增

      Linux 缺页中断

    1. To explore is to go outside one’s own knowledge, geography, or culture. To exploit is to make full use of one’s resources or to take advantage of opportunities.

      exploit make full use of one's resources to take advantage of ...

      explore is go outside one's knowledge...

    2. Explore has many connotations having to do with searching and finding out

      explore: explore new opportunity exploit: Exploit good weather and swim today. exploit miner

  9. Sep 2017
    1. 全连接层(fully connected layers,FC)在整个卷积神经网络中起到“分类器”的作用。如果说卷积层、池化层和激活函数层等操作是将原始数据映射到隐层特征空间的话,全连接层则起到将学到的“分布式特征表示”映射到样本标记空间的作用。在实际使用中,全连接层可由卷积操作实现:对前层是全连接的全连接层可以转化为卷积核为1x1的卷积;而前层是卷积层的全连接层可以转化为卷积核为hxw的全局卷积,h和w分别为前层卷积结果的高和宽(注1)。

      FC layer/ pool layer ...作用

    1. LDA的思想还是很简单的:给定训练样本,设法将样本投影到一条直线或者一张超平面上,使得同类样例点的投影尽可能的接近,异类样本点的投影尽可能远离;在对新的样本点进行分类时,将其投影到上述确定的这条直线上,再根据投影点的位置来确定新样本的类别,如下图所示(为方便可视化,以二维数据为例)

      LDA Linear Discriminant analysis finally reached this algorithm

    1. 其中wi(i=1,2,...m)为拉格朗日乘子。由于原始问题满足凸优化理论中的KKT条件,因此原始问题的解和对偶问题的解是一致的。这样我们的损失函数的优化变成了拉格朗日对偶问题的优化。

      KKT条件又在这里体现

    2. 仿射函数

      作者:Cascade 链接:https://www.zhihu.com/question/20666664/answer/15790507 来源:知乎 著作权归作者所有。商业转载请联系作者获得授权,非商业转载请注明出处。

      以下内容仅涉及图形变换,未考虑更为抽象的概念。为了方便起见,以下叙述均采用平面直角坐标系。一个矢量(1,2)可以表示为从原点指向该点的箭头。你可以对这个矢量进行缩放,比如放大两倍就变成了(2,4)这个操作可以表示为2 x(1,2)。也就是说放大k倍就是k(x,y)上面的例子写成矩阵的话就是,这里用到了矩阵乘法。<img src="https://pic2.zhimg.com/b7962fe4cfb1a49a4fda71d2b93657e9_b.jpg" data-rawwidth="98" data-rawheight="51" class="content_image" width="98">这个很简单。你也可以把矩阵中的两个值弄成不一样的。那么如果你对一张图片操作的话,横竖两个方向上的缩放倍数不同图像就变形了。方的变成长方的。这个很简单。你也可以把矩阵中的两个值弄成不一样的。那么如果你对一张图片操作的话,横竖两个方向上的缩放倍数不同图像就变形了。方的变成长方的。你也可以对矢量进行旋转。比如想把向量(1,0)逆时针旋转45度。旋转以后的向量和这个向量会构成一个三角形。旋转以后的是斜边,长度和原来向量长度一样。用勾股定理计算一下。三角形的顶点会变成(√2/2, √2/2)。这个看起来比较麻烦。但是如果你明白矩阵乘法是怎么算的,那很容易理解为什么一个旋转矩阵会是这样的:<img src="https://pic4.zhimg.com/9c70a69a8f0b3dcebc73b3aeccb2a677_b.jpg" data-rawwidth="159" data-rawheight="56" class="content_image" width="159">有些变换,比如反射。相当于你在第一种情况里面对角线上的两个值有一个是负的。那么对应的就会把这个轴翻转过去。别的都很好理解。这些变换被称为线性变换。它提供了把一个图像扭成任意形状的方法。但在二维坐标系内,用2x2的矩阵所不能表示的变换就是平移操作。你在上面所有的操作无非都是给向量的两个分量乘一个系数。没办法再加一个数。想要表达这种计算就得给你的矩阵变成这样:<img src="https://pic4.zhimg.com/6f0b8ce45cb1d2f89083f9e39d8311d7_b.jpg" data-rawwidth="95" data-rawheight="64" class="content_image" width="95">这样的话你的(x,y)向量就没法乘进去了。你可以在后面添个1,编程(x,y,1)这样的。那么变换以后的结果就是(xa1+yb1+c1,xa2+yb2+c2,1),去掉最后面的1,前面的就是线性变换加上一个平移变换的结果。这就是仿射变换。简单的说就是一个线性变换加上平移。写公式和矩阵真坑爹啊。。。

    3. 最大熵模型假设分类模型是一个条件概率分布P(Y|X), X为特征,Y为输出。给定一个训练集,(x(1),y(1)),(x(2),y(2)),...,(x(m),y(m)),其中x为n维特征向量,y为类别输出。我们的目标就是用最大熵模型选择一个最好的分类类型。

      最大熵模型

    1. 特异性(specificity)的定义图上没有直接写明,这里给出,是红色半圆除以右边的长方形。严格的数学定义如下:S=FP/(FP+TN) 

      病人为例:这里正确识别了没有生病人的比例

    1. 坐标下降法坐标轴下降法顾名思义,是沿着坐标轴的方向去下降,这和梯度下降不同。梯度下降是沿着梯度的负方向下降。不过梯度下降和坐标轴下降的共性就都是迭代法,通过启发式的方式一步步迭代求解函数的最小值。坐标轴下降法的数学依据主要是这个结论(此处不做证明):一个可微的凸函数J(θ), 其中θ是nx1的向量,即有n个维度。如果在某一点θ¯,使得J(θ)在每一个坐标轴θ¯i(i = 1,2,...n)上都是最小值,那么J(θ¯i)就是一个全局的最小值

      coordinate descent definition

    2. 就是它的损失函数不是连续可导的

      why lost function is so important? for a perfect matching model, the output of loss function should be zero. at least we want the loss function produces minimal result. this is our goal. Stochastic descending / least square / LGBS? all for this purpose.

    1. Here t {\displaystyle t} is a prespecified free parameter that determines the amount of regularisation

      prespecified free parameter determinated the amount of regularisation

  10. Aug 2017
    1. 连续可微,利用高中的知识,对求导数,然后令导数为0,就可解出最优解

      极小值. 如果有多信导数为0处呢? 需要比较?

    1. # 可以这样想:(1) 我们的两个任务:①对参数最小化L(解SVM要求),②对乘子又要最大化(拉格朗日乘子法要求), (2) 如果上面的约束条件成立,整个求和都是非负的,很好L是可以求最小值的;(3) 约束条件不成立,又要对乘子最大化,全身非负的L直接原地爆炸

      约束条件这里不是很明白?

    2. 什么是SVM?Support Vector Machine, 一个普通的SVM就是一条直线罢了,用来完美划分linearly separable的两类。但这又不是一条普通的直线,这是无数条可以分类的直线当中最完美的,因为它恰好在两个类的中间,距离两个类的点都一样远。而所谓的Support vector就是这些离分界线最近的『点』。

      这个SVM定义直观

    1.   在给出几何间隔的定义之前,咱们首先来看下,如上图所示,对于一个点 x ,令其垂直投影到超平面上的对应的为 x0 ,由于 w 是垂直于超平面的一个向量,为样本x到分类间隔的距离,我们有

      下面公式对\frac{\omega}{\left | \omega \right |} 里面的\omega 可转置 \omega\cdot \omega^{T} = (\left | \omega \right |)^2

    2. 当我们要判别一个新来的特征属于哪个类时,只需求,若大于0.5就是y=1的类,反之属于y=0类。

      threshold = 0.5 这里太不严谨了 ....

    1. How likely is this concern to happen?

      how likely is that you r right? how likely is you are right? how likely you are right? how likely is this to happen? how likely is he to go off?

    1. probe PROBEPOINT [, PROBEPOINT] { [STMT ...] }Events are specified in a special syntax calledprobe points. There are several varieties of probe points definedby the translator, and tapset scripts may define others using aliases. The provided probe points are listedin thestapprobes(3),tapset::*(3stap), andprobe::*(3stap)man pages. The STMT statement blockis executed whenever ıany of the named PROBEPOINT events occurs

      Find the probe types from 'man stapprobes'

  11. Jul 2017
    1. Suppose one person attempts to buy every single thing he has ever wanted; one puts it all in the bank and uses the money only sparingly, for special occasions; and one gives it all to charity.

      Use the money sparingly

      Use the data sparingly

    1. "American laws and American policy view the content of communications as the most private and the most valuable, but that is backwards today," said Marc Rotenberg, the executive director of the Electronic Privacy Information Center, a Washington group.

      But that is backwards today

      another,, go backwards

    2. The system is backwards compatible with existing Playstation games, while Dolphin will mark Nintendo's switch from cartridges to DVDs.

      backwards: be backwards compatible with ...

      this feature is backwards (落后的)

  12. www.thefreedictionary.com www.thefreedictionary.com
    1. clut•ter (ˈklʌt ər) v.t. 1. to fill or litter with things in a disorderly manner: Newspapers cluttered the living room. v.i. 2. Dial. to bustle. n. 3. a disorderly heap or assemblage; litter. 4. a confused state. 5. echoes on a radar screen that do not come from the target.

      clutter -> Radar screen

      background clutter

    1. is aware that certain conduct may amount to a crime for which they will be held accountable.

      amount to = lead to

      What this amount to is that ''' What this represent ..."

      this amount to nothing

  13. www.thefreedictionary.com www.thefreedictionary.com
    1. no·ta·tion  (nō-tā′shən)n.1. a. A system of figures or symbols used in a specialized field to represent numbers, quantities, tones, or values: musical notation.b. The act or process of using such a system.2. A brief note; an annotation: marginal notations.

      a system of figures or symbols used in specialized filed to represent numbers ....

      Familiarize yourself with notations ...

    1. New York's budget has been overdue now for weeks (it was due April 1) but at that time, the legislature will set what they feel is appropriate SUNY tuition.

      it was due Apri 1 it is due tomorrow it was due this month it is due Jan 20. 'with day/time directly' ''has been overdue for weeks"

    1. Orientation tuning, but not direction selectivity, is invariant to temporal frequency in primary visual cortex

      is invariant to temporal frequency

      is invariant to stretching

      is invariant to age ( is not related to ago, not associated with ago).... be invariant to sth. 与...无关

    1. 1.在同一个意群中,相邻的两个词,前者以辅音音素结尾,后者以元音音素开头,往往要拼在一起连读。如:     He is a student.(is与a要连读)     That is a right answer.(That与is,is和a,right和answer都可以连读)     I’ll be back in half an hour.(back和in,half和an,an与hour都可以连读)。

      辅音+元音