python 读写数据

开源标准数据集 —— mnist（手写字符识别）

１. 使用 python 读取和解析 mnist.pkl.gz

import pickle
import gzip
from PIL import Image

def load_data():
with gzip.open('./mnist.pkl.gz') as fp:
training_data, valid_data, test_data = pickle.load(fp)
return training_data, valid_data, test_data

training_data, valid_data, test_data = load_data()
print len(training_data[0])
print len(valid_data[0])
print len(test_data[0])
print len(training_data[0][0])

I = training_data[0][0]
I.resize((28, 28))
im = Image.fromarray((I*256).astype('uint8'))
im.show()
im.save('5.png')

def normalize(darr, eps=1e-8):
# normalize(x) = (x-min)/(max-min)
darr -= darr.min()
darr *= 1./(darr.max()+eps)
return darr

def tile_raster_images(X, image_shape, tile_shape,
tile_spacing=(0, 0), normalize_rows=True, output_pixel_vals=True):
# image_shape：每一个砖的高和宽，
# tile_shape：在横纵两个方向上分别有多少砖
# tile_spacing：砖与砖之间的距离
# normalize_rows：是否对砖进行归一化
# output_pixel_vals：是否对砖以图像的形式进行显示

assert len(image_shape) == 2
assert len(tile_shape) == 2
assert len(tile_spacing) == 2
# 对参数进行断言，确保它们都是二维元组
output_shape = [
(ishp + tsp)*tshp-tsp
for ishp, tshp, tsp in zip(image_shape, tile_shape, tile_spacing)
]
# image_shape == (28, 28)   mnist data
# tile_shape == (10, 10), tile_spacing == (1, 1)
# [(28+1)*10-1]*[(28+1)*10-1]

H, W = image_shape
Hs, Ws = tile_spacing
dt = 'uint8' if output_pixel_vals else X.dtype
# python 风格的三目运算符
output_array = numpy.zeros(output_shape, dtype=dt)

# 开始贴砖
for i in range(tile_shape[0]):
for j in range(tile_shape[1]):
if i*tile_shape[1]+j < X.shape[0]:
# X的每一行是一个图像（二维）flatten后的（一维的行向量）
this_x = X[i*tile_shape[1]+j]
this_image = normalize(this_x.reshape(image_shape)) if normalize_rows else this_x.reshape(image_shape)
c = 255 if output_pixel_vals else 1
output_array[
i*(H+Hs):i*(H+Hs)+H, j*(W+Ws):j*(W+Ws)+W
] = this_image*c
return output_array

import numpy
from PIL import Image

X = numpy.random.randn(500, 28*28)
arr = tile_raster_images(X, image_shape=(28, 28),
tile_shape=(12, 12), tile_spacing=(1, 1))
img = Image.fromarray(arr)
img.show()
img.save('./砖块可视化.png')
# 这里也可使用 matplotlib 进行显示
# plt.imshow(img, cmap='gray')
# plt.show()

#this is a comment

'''
ABC
ABC
ABC
'''
"""
ABC
ABC
ABC
"""

#coding=utf-8

# -*- coding: utf-8 -*-

#!/usr/bin/python

# -*- coding: utf-8 -*-