2017年06月_一呆飞仙

12月 11月 10月 09月 08月 07月 06月 05月 04月 03月 02月 01月

原创 numpy.transpose难点解析

先来看一个例子import numpy as nparr1 = np.array(range(1, 37))print 'arr1:','\n',arr1arr2 = arr1.reshape(4, 3, 3)print 'arr2:','\n',arr2arr3 = arr2.transpose(1，0，2)print 'arr3:','\n',arr3arr4 = arr...

2017-06-27 14:19:21 545

原创 theano-xnor-net代码注释9 pylearn2/cifar10.py

""".. todo:: WRITEME"""import osimport loggingimport numpyfrom theano.compat.six.moves import xrangefrom pylearn2.datasets import cache, dense_design_matrixfrom pylearn2.expr.preprocessing imp

2017-06-20 10:48:27 971

原创 theano-xnor-net代码注释8 xnornet_layers.py

""" Class and method definition for the layers in XNOR-Net"""import theanoimport theano.tensor.nnetimport numpy as npimport lasagneimport theano.tensor as Timport timefrom fxp_helper import to_

2017-06-20 10:43:05 885

原创 theano-xnor-net代码注释7 inf_layers.py

import theanoimport theano.tensor as Timport numpy as npimport lasagnefrom fxp_helper import to_fixed_point_theano# Classic batch normalization layer with fixed point simulation feature for the out

2017-06-20 10:41:21 550

原创 theano-xnor-net代码注释6 fxp_helper.py

import numpy as npimport theanoimport theano.tensor as Timport math#to_fixed_point_theano函数就是设定一个阈值，即max=data_bits能表达最大的无符号整数,阈值为[-max，max]，中间还存在一些扩大与近似操作#最后的目的就是加速收敛def to_fixed_point_theano(inp

2017-06-20 10:39:43 385

原创 theano-xnor-net代码注释5 cifar10_test.py

import sys, os, timeimport argparseimport lasagneimport numpy as npimport theanoimport theano.tensor as Timport cPickle, timeimport inf_layersfrom fxp_helper import convert_fxp_format, fixed_po

2017-06-20 10:38:20 771

原创 theano-xnor-net代码注释4 bnn_utils.py

# This code is taken from the BinaryNet implementation by Matthieu Courbariaux# The original code can be found here https://github.com/MatthieuCourbariaux/BinaryNet# The LICENSE for this piece of cod

2017-06-20 10:34:23 1662 4

原创 theano-xnor-net代码注释3 xnor_net.py

# *_*coding:utf-8 *_*""" Class and method definition for the layers in XNOR-Net(用于构建神经网络各层的类和函数)"""import theanoimport theano.tensor.nnetimport numpy as npimport lasagneimport theano.tensor as T

2017-06-20 10:32:45 985

原创 theano-xnor-net代码注释2 cnn_utils.py

import numpy as npimport cPickleimport cv2from pylearn2.datasets.mnist import MNISTfrom pylearn2.datasets.cifar10 import CIFAR10from pylearn2.datasets.svhn import SVHN_DATASET_SIZE = {'mnist' : 50

2017-06-20 10:30:40 1127

原创 theano-xnor-net代码注释 cifar10_train.py

# *_*coding:utf-8 *_*import sys, os, timeimport lasagneimport numpy as npimport theanoimport theano.tensor as Timport cPickleimport xnor_netimport cnn_utilsfrom external import bnn_utilsimport

2017-06-20 10:27:12 1681 5

本文转载自：http://blog.csdn.net/zhuiqiuk/article/details/53613879https://handong1587.github.io/deep_learning/2015/10/09/nlp.htmlJump to...LeaderboardPapersR-CNNMultiBoxSPP-NetDeepID-NetNoCFast R-CNNDeepBox...

2017-06-12 09:24:55 2486 1

base64工具

本压缩文件中包含base64.h与base64.cpp两个文件，只需将该文件放入visual studio工程头文件中，即可实现base64编解码

2017-12-26

用自己的数据制作python版本cifar10数据集

用自己的数据制作python版本cifar10数据集，该版本为替换版本，替换标签为9的数据。需要准备6000张图片，替换整合为6个二进制cifar10数据集

2017-05-03

用自己的数据制作cifar10数据集

2017-05-03

NCVPixelOperations.hpp

对于opencv配置cuda出现的 error: a storage class is not allowed in an explicit specialization问题

2016-11-24

空空如也

TA创建的收藏夹 TA关注的收藏夹

TA关注的人

一呆飞仙的博客