CNN_Structure_2

这篇文章还是一篇在学习中整理的笔记,不想做优化,就是这么任性

CNN如何做到的可视化?

在第一层,将权重可视化,因为他们直接和图像相关。那我们如何将后面的卷积层可视化?这些可视化方法都是对神经元的反应做出的可视化,而不是对滤波器本身,实际上目前并没有好的方法对滤波器本身做可视化,所以在ConventnetJS中你可以找到权重,但是并不能直接对他们做出很好的解释。因为权重自己没有很好的解释意义

Pooling Layer 的真正目的以及这么处理的原因

More

MapReduce VS RDBMs

Now, the RDBMs – Rational Database Management Systems, is good for updating a small portion of a big database. RDBMs uses a traditional B-Tree, which is highly dependent on the time required to perform seek operations.

Compared to this, MapReduce is good for updating all, or a majority of a big database. MapReduce uses sort and merge to rebuild the database which depends more on transfer operations.

RDBMs is good for applications that required the data sets if the database to be very frequently updated such as in point queries or small dataset updates. MapReduce is better for WORM, which is a write once and read many times based data applications.

More

Cloud Service Models

We are going to look into further details of the cloud service models.

First of all, look at this picture. At the bottom which is the infrastructure, the IaaS. And in here, virtual machines, server storage, network, is basically what is provides in the infrastructure. Above that is PaaS, where this is a platform as a service, and it includes database, web server and deployment tools. The next one on the top is SaaS and this is software as a Service. Here, CRM, email, games, virtual desktop and what is actually in the software application domain.

More

Cloud Introduction

At first, what does cloud computing do?

Well. It provides online data storage. It enables configuration and accessing of online applications. Apart from that, it provides a variety of software usage. And also it provides computing platform and computing infrastructure.

Now, that may not have lightened up what the meaning of cloud computing is, so here comes an example.

More

中断以及中断处理

中断

事实上所有计算机都提供了允许其他模块(I/O、存储器)中断处理器正常处理过程的机制。分类如下:

  • 程序中断:在某些条件下由指令执行的结果产生,例如算术溢出、除数为0、试图执行一条非法的机器指令以及访问到用户不允许的存储器位置。
  • 时钟中断:由处理器内部的计时器产生,允许操作系统以一定规律执行函数
  • I/O中断:由I/O控制器产生,用于发信号通知一个操作的正常完成或各种错误条件
  • 硬件故障中断:由诸如掉电或存储器奇偶错误之类的故障产生

More

MapReduce & Hadoop

Hadoop 是什么?Hadoop 又与 MapReduce 有什么关系?本篇文章会简单介绍两者之间的关系,算是自己在学习过程中做的笔记吧。

我们在练习过程中可能就使用几M的数据量,但是真正在实践中我们接触的数据可比这大得多,这是如果还使用一台电脑进行运算的话就显得十分笨拙,并且实际上远远超出了我们的计算能力。幸运的是一些开源的软件项目提供了处理海量数据的方案,其中一个项目就是 Hadoop,它采用JAVA语言编写,支持在大量机器上分布处理数据。

Hadoop 是 MapReduce 框架的一个开源实现。本篇笔记简单介绍 MapReduce 和 Hadoop 项目。

More