2016年01月_Firehotest

原创 Self summary: What does typical CNN do and What is Soft-max?

Here is the image stoled from guide of torch: https://github.com/soumith/cvpr2015/blob/master/Deep%20Learning%20with%20Torch.ipynbFirst, we must know what is convolution.Here is a

2016-01-31 17:23:49 678

转载 Study Notes: Lua language for Torch

Lua Data types: Lua has two data structure built-in, a table: {} and tensor(matrix). Also there are Strings, numbers. String use '' : a = 'hello'# is the length operator in Lua

2016-01-28 19:05:29 1400

转载 Linux: gcc and gcc -O2, configure, make, make install and makefile, sed

GCC: Gcc is a cross-platform complier. It can be run on windows and linux and so on. As it has been developed well, it can compile many language now. If you need to use gcc to compile, onc

2016-01-28 10:31:43 1130

转载 Linux Command: mkdir -p; find; gitclone; hardconnection and softconnection

1/ mkdir的参数：-m, --mode=模式，设定权限 (类似 chmod)，而不是 rwxrwxrwx 减 umask-p, --parents 可以是一个路径名称。此时若路径中的某些目录尚不存在,加上此选项后,系统将自动建立好那些尚不存在的目录,即一次可以建立多个目录; -v, --verbose 每次创建新目录都显示信息 --help 显示此帮助信息并退出

2016-01-27 18:31:42 604

原创 Study Note: Logistic Regression(2)

First of all, let's add something missed from the previous section:What is the practical meaning of output of sigmoid function?It is the posterior possibility of y = 1. P(y=1 |x,Θ) = ? If

2016-01-24 12:37:05 1264

原创 Study Note: Logistic Regression(1)

Logistic regression is created for logistic problem. Therefore, the output of such regression should be binary value. Then here is the question, how to create a regression function whose output

2016-01-23 20:20:12 558

原创 Study Note: Multiple Variables Regression

How does the bia unit of neural network come out?Let's assume the parameters matrix(vector) is Θ = In order to let variables matrix(vector) is suitable for multiply, we make a extra = 1 f

2016-01-23 14:40:57 606

原创 Study Notes of Neural Network Optimization(1)

1/ Things to do before train the network: pre-train the networkNormal network pre-training methods include using stacks of RBMs (Restrict Boltzmann Machine); autoencoders and Deep Boltzmann Mach

2016-01-21 09:52:36 772

原创 Linux to read the Processor model, physical cores, virtual cores and L1, L2 cache

The information about the CPU of a Linux computer is stored in the file /proc/cpuinfo(cat /proc/cpuinfo)It is a long file so that we need to use pipeline tool to process our information. I

2016-01-20 10:22:04 627

原创 A summary of dropout technique

1/ Motivation:Typically, for overcoming overfitting, we introduce methods like stopping training as soon as performance on a validation set starts to get worse; soft weight sharing[2] and regula

2016-01-19 14:28:22 799

原创 A summary about gradient descent

There are three kind of gradient descent now:1/ Traditional:Like a people who wants to get down of the hill as soon as possible. Then he/she will find a way that has a big steepness of loc

2016-01-19 11:03:28 892

转载 Linux source command & bashrc/ profile

Source command is used to compile some critical instructions. Like the instructions in bashrc or profile. The bashrc and profile files are used to run some important instructions once you open y

2016-01-18 23:38:22 476

原创 Study Note: About CNN

1/ The difference of CNN and ordinary Neural networkAns: Typical neural network is fully connected network and each unit's activation of this network is:The reason for choosing thi

2016-01-15 11:18:35 835

Steven Li's Zone