{"id":944773,"date":"2024-12-26T23:12:52","date_gmt":"2024-12-26T15:12:52","guid":{"rendered":""},"modified":"2024-12-26T23:12:54","modified_gmt":"2024-12-26T15:12:54","slug":"python%e5%a6%82%e4%bd%95%e8%af%bb%e5%8f%96mnist","status":"publish","type":"post","link":"https:\/\/docs.pingcode.com\/ask\/944773.html","title":{"rendered":"python\u5982\u4f55\u8bfb\u53d6mnist"},"content":{"rendered":"

\"python\u5982\u4f55\u8bfb\u53d6mnist\"<\/p>\n

Python\u8bfb\u53d6MNIST\u6570\u636e\u96c6\u7684\u65b9\u6cd5\u4e3b\u8981\u5305\u62ec\uff1a\u4f7f\u7528Keras\u5e93\u3001\u4f7f\u7528TensorFlow\u5e93\u3001\u4f7f\u7528Scikit-learn\u5e93\u3002<\/strong>\u5176\u4e2d\uff0c\u4f7f\u7528Keras\u5e93\u662f\u6700\u5e38\u89c1\u548c\u7b80\u4fbf\u7684\u65b9\u6cd5\uff0c\u56e0\u4e3aKeras\u5df2\u7ecf\u96c6\u6210\u4e86MNIST\u6570\u636e\u96c6\u3002\u4e0b\u9762\u5c06\u8be6\u7ec6\u4ecb\u7ecd\u4f7f\u7528Keras\u5e93\u8bfb\u53d6MNIST\u6570\u636e\u96c6\u7684\u6b65\u9aa4\u3002<\/p>\n<\/p>\n

\u4e00\u3001\u4f7f\u7528Keras\u8bfb\u53d6MNIST\u6570\u636e\u96c6<\/h3>\n<\/p>\n

Keras\u662f\u4e00\u4e2a\u9ad8\u7ea7\u795e\u7ecf\u7f51\u7edcAPI\uff0c\u8fd0\u884c\u5728TensorFlow\u4e4b\u4e0a\uff0c\u63d0\u4f9b\u4e86\u7b80\u5355\u7684\u65b9\u5f0f\u6765\u8bbf\u95eeMNIST\u6570\u636e\u96c6\u3002\u4f7f\u7528Keras\u8bfb\u53d6MNIST\u6570\u636e\u96c6\uff0c\u53ea\u9700\u51e0\u884c\u4ee3\u7801\u5373\u53ef\u5b8c\u6210\u3002<\/p>\n<\/p>\n

1. \u5bfc\u5165\u6240\u9700\u7684\u5e93<\/h4>\n<\/p>\n

\u5728\u5f00\u59cb\u4e4b\u524d\uff0c\u6211\u4eec\u9700\u8981\u5bfc\u5165\u5fc5\u8981\u7684Python\u5e93\u3002\u9664\u4e86Keras\u672c\u8eab\uff0c\u8fd8\u9700\u8981\u4f7f\u7528Numpy\u6765\u5904\u7406\u6570\u7ec4\u3002<\/p>\n<\/p>\n

import numpy as np<\/p>\n
from keras.datasets import mnist<\/p>\n
<\/code><\/pre>\n<\/p>\n
2. \u52a0\u8f7dMNIST\u6570\u636e\u96c6<\/h4>\n<\/p>\n
Keras\u63d0\u4f9b\u4e86\u4e00\u4e2a\u975e\u5e38\u7b80\u5355\u7684\u63a5\u53e3\u6765\u52a0\u8f7dMNIST\u6570\u636e\u96c6\uff0c\u53ea\u9700\u8c03\u7528mnist.load_data()<\/code>\u51fd\u6570\u5373\u53ef\u3002\u8be5\u51fd\u6570\u4f1a\u8fd4\u56de\u4e24\u4e2a\u5143\u7ec4\uff1a\u8bad\u7ec3\u6570\u636e\u548c\u6d4b\u8bd5\u6570\u636e\u3002<\/p>\n<\/p>\n
(x_trAI<\/a>n, y_train), (x_test, y_test) = mnist.load_data()<\/p>\n
<\/code><\/pre>\n<\/p>\n
3. \u6570\u636e\u9884\u5904\u7406<\/h4>\n<\/p>\n
\u5728\u52a0\u8f7d\u6570\u636e\u4e4b\u540e\uff0c\u901a\u5e38\u9700\u8981\u5bf9\u6570\u636e\u8fdb\u884c\u9884\u5904\u7406\u3002MNIST\u6570\u636e\u96c6\u4e2d\u7684\u56fe\u50cf\u662f\u7070\u5ea6\u56fe\u50cf\uff0c\u50cf\u7d20\u503c\u57280\u5230255\u4e4b\u95f4\u3002\u4e3a\u4e86\u4f7f\u6a21\u578b\u66f4\u5bb9\u6613\u8bad\u7ec3\uff0c\u6211\u4eec\u901a\u5e38\u5c06\u50cf\u7d20\u503c\u5f52\u4e00\u5316\u52300\u52301\u4e4b\u95f4\u3002<\/p>\n<\/p>\n
x_train = x_train.astype('float32') \/ 255<\/p>\n
x_test = x_test.astype('float32') \/ 255<\/p>\n
<\/code><\/pre>\n<\/p>\n
4. \u6570\u636e\u5f62\u72b6\u8c03\u6574<\/h4>\n<\/p>\n
MNIST\u6570\u636e\u96c6\u4e2d\u7684\u6bcf\u4e2a\u56fe\u50cf\u662f28×28\u50cf\u7d20\u7684\u4e8c\u7ef4\u6570\u7ec4\u3002\u5728\u4f7f\u7528\u795e\u7ecf\u7f51\u7edc\u65f6\uff0c\u6211\u4eec\u901a\u5e38\u9700\u8981\u5c06\u5176\u8f6c\u6362\u4e3a\u4e00\u7ef4\u6570\u7ec4\u3002<\/p>\n<\/p>\n
x_train = x_train.reshape(x_train.shape[0], 28, 28, 1)<\/p>\n
x_test = x_test.reshape(x_test.shape[0], 28, 28, 1)<\/p>\n
<\/code><\/pre>\n<\/p>\n
5. \u6807\u7b7e\u7f16\u7801<\/h4>\n<\/p>\n
MNIST\u6570\u636e\u96c6\u7684\u6807\u7b7e\u662f\u6570\u5b57\u5f62\u5f0f\u7684\u3002\u4e3a\u4e86\u4f7f\u7528\u795e\u7ecf\u7f51\u7edc\u8fdb\u884c\u8bad\u7ec3\uff0c\u6211\u4eec\u9700\u8981\u5c06\u8fd9\u4e9b\u6807\u7b7e\u8fdb\u884cone-hot\u7f16\u7801\u3002<\/p>\n<\/p>\n
from keras.utils import to_categorical<\/p>\n
y_train = to_categorical(y_train, 10)<\/p>\n
y_test = to_categorical(y_test, 10)<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u4e8c\u3001\u4f7f\u7528TensorFlow\u8bfb\u53d6MNIST\u6570\u636e\u96c6<\/h3>\n<\/p>\n
\u9664\u4e86Keras\uff0cTensorFlow\u4e5f\u63d0\u4f9b\u4e86\u8bfb\u53d6MNIST\u6570\u636e\u96c6\u7684\u529f\u80fd\u3002\u4f7f\u7528TensorFlow\u8bfb\u53d6MNIST\u6570\u636e\u96c6\u7684\u6b65\u9aa4\u5982\u4e0b\u3002<\/p>\n<\/p>\n
1. \u5bfc\u5165TensorFlow\u5e93<\/h4>\n<\/p>\n
\u5728\u5f00\u59cb\u4e4b\u524d\uff0c\u6211\u4eec\u9700\u8981\u5bfc\u5165TensorFlow\u5e93\u3002<\/p>\n<\/p>\n
import tensorflow as tf<\/p>\n
<\/code><\/pre>\n<\/p>\n
2. \u52a0\u8f7dMNIST\u6570\u636e\u96c6<\/h4>\n<\/p>\n
\u4f7f\u7528TensorFlow\u8bfb\u53d6MNIST\u6570\u636e\u96c6\uff0c\u53ef\u4ee5\u901a\u8fc7tf.keras.datasets.mnist<\/code>\u6a21\u5757\u5b9e\u73b0\u3002<\/p>\n<\/p>\n
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()<\/p>\n
<\/code><\/pre>\n<\/p>\n
3. \u6570\u636e\u9884\u5904\u7406<\/h4>\n<\/p>\n
\u4e0eKeras\u7c7b\u4f3c\uff0c\u6211\u4eec\u4e5f\u9700\u8981\u5bf9\u6570\u636e\u8fdb\u884c\u9884\u5904\u7406\u3002\u5c06\u50cf\u7d20\u503c\u5f52\u4e00\u5316\u52300\u52301\u4e4b\u95f4\u3002<\/p>\n<\/p>\n
x_train = x_train \/ 255.0<\/p>\n
x_test = x_test \/ 255.0<\/p>\n
<\/code><\/pre>\n<\/p>\n
4. \u6807\u7b7e\u7f16\u7801<\/h4>\n<\/p>\n
\u4f7f\u7528TensorFlow\u8fdb\u884c\u6807\u7b7e\u7f16\u7801\u4e0eKeras\u7c7b\u4f3c\uff0c\u4e5f\u9700\u8981\u5c06\u6807\u7b7e\u8fdb\u884cone-hot\u7f16\u7801\u3002<\/p>\n<\/p>\n
y_train = tf.keras.utils.to_categorical(y_train, 10)<\/p>\n
y_test = tf.keras.utils.to_categorical(y_test, 10)<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u4e09\u3001\u4f7f\u7528Scikit-learn\u8bfb\u53d6MNIST\u6570\u636e\u96c6<\/h3>\n<\/p>\n
Scikit-learn\u662f\u4e00\u4e2a\u5f3a\u5927\u7684\u673a\u5668\u5b66\u4e60<\/a>\u5e93\uff0c\u4e5f\u53ef\u4ee5\u7528\u4e8e\u8bfb\u53d6MNIST\u6570\u636e\u96c6\u3002<\/p>\n<\/p>\n
1. \u5bfc\u5165Scikit-learn\u5e93<\/h4>\n<\/p>\n
\u6211\u4eec\u9700\u8981\u5bfc\u5165Scikit-learn\u5e93\u548cNumpy\u5e93\u3002<\/p>\n<\/p>\n
from sklearn.datasets import fetch_openml<\/p>\n
import numpy as np<\/p>\n
<\/code><\/pre>\n<\/p>\n
2. \u52a0\u8f7dMNIST\u6570\u636e\u96c6<\/h4>\n<\/p>\n
\u4f7f\u7528Scikit-learn\u7684fetch_openml<\/code>\u51fd\u6570\u53ef\u4ee5\u8f7b\u677e\u52a0\u8f7dMNIST\u6570\u636e\u96c6\u3002<\/p>\n<\/p>\n
mnist = fetch_openml('mnist_784', version=1)<\/p>\n
<\/code><\/pre>\n<\/p>\n
3. \u6570\u636e\u9884\u5904\u7406<\/h4>\n<\/p>\n
\u5728\u4f7f\u7528Scikit-learn\u65f6\uff0c\u6211\u4eec\u9700\u8981\u624b\u52a8\u8fdb\u884c\u6570\u636e\u9884\u5904\u7406\u3002\u9996\u5148\u5c06\u6570\u636e\u8f6c\u6362\u4e3anumpy\u6570\u7ec4\uff0c\u5e76\u5c06\u50cf\u7d20\u503c\u5f52\u4e00\u5316\u52300\u52301\u4e4b\u95f4\u3002<\/p>\n<\/p>\n
x = mnist.data.astype(np.float32) \/ 255<\/p>\n
y = mnist.target.astype(np.int64)<\/p>\n
<\/code><\/pre>\n<\/p>\n
4. \u6570\u636e\u62c6\u5206<\/h4>\n<\/p>\n
\u4e3a\u4e86\u8fdb\u884c\u8bad\u7ec3\u548c\u6d4b\u8bd5\uff0c\u6211\u4eec\u9700\u8981\u5c06\u6570\u636e\u96c6\u62c6\u5206\u4e3a\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u3002<\/p>\n<\/p>\n
from sklearn.model_selection import train_test_split<\/p>\n
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.2, random_state=42)<\/p>\n
<\/code><\/pre>\n<\/p>\n
5. \u6807\u7b7e\u7f16\u7801<\/h4>\n<\/p>\n
\u4e0e\u4e4b\u524d\u7c7b\u4f3c\uff0c\u9700\u8981\u5bf9\u6807\u7b7e\u8fdb\u884cone-hot\u7f16\u7801\u3002<\/p>\n<\/p>\n
from keras.utils import to_categorical<\/p>\n
y_train = to_categorical(y_train, 10)<\/p>\n
y_test = to_categorical(y_test, 10)<\/p>\n
<\/code><\/pre>\n<\/p>\n
\u56db\u3001MNIST\u6570\u636e\u96c6\u7684\u5e94\u7528<\/h3>\n<\/p>\n
MNIST\u6570\u636e\u96c6\u662f\u4e00\u4e2a\u975e\u5e38\u7ecf\u5178\u7684\u6570\u636e\u96c6\uff0c\u5e7f\u6cdb\u5e94\u7528\u4e8e\u673a\u5668\u5b66\u4e60\u548c\u6df1\u5ea6\u5b66\u4e60\u7684\u7814\u7a76\u4e2d\u3002\u5b83\u5e38\u88ab\u7528\u4e8e\u6d4b\u8bd5\u65b0\u7684\u7b97\u6cd5\u548c\u6a21\u578b\u3002\u4ee5\u4e0b\u662f\u4e00\u4e9b\u5e38\u89c1\u7684\u5e94\u7528\u573a\u666f\u3002<\/p>\n<\/p>\n
1. \u624b\u5199\u6570\u5b57\u8bc6\u522b<\/h4>\n<\/p>\n
MNIST\u6570\u636e\u96c6\u7684\u4e3b\u8981\u5e94\u7528\u573a\u666f\u662f\u624b\u5199\u6570\u5b57\u8bc6\u522b\u3002\u901a\u8fc7\u8bad\u7ec3\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\uff0c\u6211\u4eec\u53ef\u4ee5\u5b9e\u73b0\u5bf9\u624b\u5199\u6570\u5b57\u7684\u81ea\u52a8\u8bc6\u522b\u3002<\/p>\n<\/p>\n
2. \u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u6d4b\u8bd5<\/h4>\n<\/p>\n
MNIST\u6570\u636e\u96c6\u7531\u4e8e\u5176\u7b80\u5355\u6027\u548c\u6613\u7528\u6027\uff0c\u5e38\u7528\u4e8e\u6d4b\u8bd5\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u3002\u7814\u7a76\u4eba\u5458\u53ef\u4ee5\u5feb\u901f\u9a8c\u8bc1\u6a21\u578b\u7684\u6548\u679c\u3002<\/p>\n<\/p>\n
3. \u6559\u5b66\u548c\u5b66\u4e60<\/h4>\n<\/p>\n
MNIST\u6570\u636e\u96c6\u4e5f\u662f\u673a\u5668\u5b66\u4e60\u548c\u6df1\u5ea6\u5b66\u4e60\u6559\u5b66\u4e2d\u5e38\u7528\u7684\u6570\u636e\u96c6\u3002\u5b83\u7b80\u5355\u6613\u61c2\uff0c\u9002\u5408\u521d\u5b66\u8005\u8fdb\u884c\u5b66\u4e60\u548c\u5b9e\u8df5\u3002<\/p>\n<\/p>\n
\u4e94\u3001\u603b\u7ed3<\/h3>\n<\/p>\n
\u5728Python\u4e2d\u8bfb\u53d6MNIST\u6570\u636e\u96c6\u6709\u591a\u79cd\u65b9\u6cd5\uff0c\u5305\u62ec\u4f7f\u7528Keras\u3001TensorFlow\u548cScikit-learn\u3002\u8fd9\u4e9b\u65b9\u6cd5\u5404\u6709\u4f18\u7f3a\u70b9\uff0cKeras\u548cTensorFlow\u63d0\u4f9b\u4e86\u7b80\u5355\u7684\u63a5\u53e3\uff0c\u975e\u5e38\u9002\u5408\u521d\u5b66\u8005\uff1bScikit-learn\u63d0\u4f9b\u4e86\u66f4\u591a\u7684\u7075\u6d3b\u6027\uff0c\u9002\u5408\u9700\u8981\u81ea\u5b9a\u4e49\u5904\u7406\u6d41\u7a0b\u7684\u9ad8\u7ea7\u7528\u6237\u3002\u65e0\u8bba\u9009\u62e9\u54ea\u79cd\u65b9\u6cd5\uff0c\u90fd\u9700\u8981\u5bf9\u6570\u636e\u8fdb\u884c\u9884\u5904\u7406\uff0c\u5305\u62ec\u50cf\u7d20\u503c\u5f52\u4e00\u5316\u548c\u6807\u7b7e\u7f16\u7801\u3002MNIST\u6570\u636e\u96c6\u5e7f\u6cdb\u5e94\u7528\u4e8e\u673a\u5668\u5b66\u4e60\u548c\u6df1\u5ea6\u5b66\u4e60\u7684\u7814\u7a76\u4e2d\uff0c\u662f\u6d4b\u8bd5\u65b0\u7b97\u6cd5\u548c\u6a21\u578b\u7684\u7406\u60f3\u9009\u62e9\u3002<\/p>\n<\/p>\n
\u76f8\u5173\u95ee\u7b54FAQs\uff1a<\/strong><\/h2>\n
 \u5982\u4f55\u4f7f\u7528Python\u8bfb\u53d6MNIST\u6570\u636e\u96c6\uff1f<\/strong>
Python\u63d0\u4f9b\u4e86\u591a\u79cd\u5e93\u6765\u8bfb\u53d6MNIST\u6570\u636e\u96c6\uff0c\u5176\u4e2d\u6700\u5e38\u7528\u7684\u662ftensorflow<\/code>\u548ckeras<\/code>\u3002\u901a\u8fc7\u8fd9\u4e9b\u5e93\uff0c\u7528\u6237\u53ef\u4ee5\u65b9\u4fbf\u5730\u4e0b\u8f7d\u548c\u52a0\u8f7dMNIST\u6570\u636e\u96c6\uff0c\u4f8b\u5982\u4f7f\u7528keras.datasets.mnist.load_data()<\/code>\u51fd\u6570\uff0c\u8be5\u51fd\u6570\u4f1a\u81ea\u52a8\u4e0b\u8f7d\u6570\u636e\u5e76\u8fd4\u56de\u8bad\u7ec3\u548c\u6d4b\u8bd5\u6570\u636e\u96c6\u7684\u5143\u7ec4\u3002\u8fd9\u4f7f\u5f97\u5904\u7406\u548c\u5206\u6790\u6570\u636e\u53d8\u5f97\u7b80\u5355\u9ad8\u6548\u3002<\/p>\n
\u8bfb\u53d6MNIST\u6570\u636e\u96c6\u65f6\u9700\u8981\u6ce8\u610f\u54ea\u4e9b\u6570\u636e\u683c\u5f0f\uff1f<\/strong>
MNIST\u6570\u636e\u96c6\u5305\u542b\u624b\u5199\u6570\u5b57\u7684\u56fe\u50cf\uff0c\u6bcf\u4e2a\u56fe\u50cf\u4e3a28×28\u50cf\u7d20\u7684\u7070\u5ea6\u56fe\u50cf\u3002\u8bfb\u53d6\u6570\u636e\u65f6\uff0c\u6ce8\u610f\u56fe\u50cf\u7684\u5f62\u72b6\u548c\u6570\u636e\u7c7b\u578b\uff0c\u901a\u5e38\u4f1a\u5c06\u56fe\u50cf\u6570\u636e\u5f52\u4e00\u5316\u52300\u52301\u4e4b\u95f4\uff0c\u4ee5\u63d0\u9ad8\u6a21\u578b\u7684\u8bad\u7ec3\u6548\u679c\u3002\u6b64\u5916\uff0c\u6807\u7b7e\u6570\u636e\u662f\u5bf9\u5e94\u7684\u6570\u5b57\uff080-9\uff09\uff0c\u5728\u8fdb\u884c\u5206\u7c7b\u4efb\u52a1\u65f6\uff0c\u786e\u4fdd\u6807\u7b7e\u4e0e\u56fe\u50cf\u6b63\u786e\u5339\u914d\u3002<\/p>\n
\u5982\u4f55\u5728Python\u4e2d\u53ef\u89c6\u5316MNIST\u6570\u636e\u96c6\u7684\u56fe\u50cf\uff1f<\/strong>
\u53ef\u89c6\u5316\u662f\u7406\u89e3\u548c\u5206\u6790\u6570\u636e\u7684\u91cd\u8981\u6b65\u9aa4\u3002\u53ef\u4ee5\u4f7f\u7528matplotlib<\/code>\u5e93\u6765\u663e\u793aMNIST\u6570\u636e\u96c6\u4e2d\u7684\u56fe\u50cf\u3002\u901a\u8fc7\u8c03\u7528plt.imshow()<\/code>\u51fd\u6570\uff0c\u7528\u6237\u53ef\u4ee5\u5c06\u56fe\u50cf\u6570\u7ec4\u4f20\u5165\u5e76\u8bbe\u7f6e\u989c\u8272\u6620\u5c04\u4e3a\u7070\u5ea6\uff0c\u4ee5\u4fbf\u66f4\u6e05\u6670\u5730\u770b\u5230\u624b\u5199\u6570\u5b57\u3002\u901a\u8fc7\u8bbe\u7f6e\u9002\u5f53\u7684\u6807\u9898\u548c\u5750\u6807\u8f74\uff0c\u7528\u6237\u53ef\u4ee5\u5bf9\u56fe\u50cf\u8fdb\u884c\u8fdb\u4e00\u6b65\u7684\u5206\u6790\u548c\u7814\u7a76\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"Python\u8bfb\u53d6MNIST\u6570\u636e\u96c6\u7684\u65b9\u6cd5\u4e3b\u8981\u5305\u62ec\uff1a\u4f7f\u7528Keras\u5e93\u3001\u4f7f\u7528TensorFlow\u5e93\u3001\u4f7f\u7528Scikit […]","protected":false},"author":3,"featured_media":944784,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[37],"tags":[],"acf":[],"_links":{"self":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/944773"}],"collection":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/comments?post=944773"}],"version-history":[{"count":"1","href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/944773\/revisions"}],"predecessor-version":[{"id":944786,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/944773\/revisions\/944786"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/media\/944784"}],"wp:attachment":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/media?parent=944773"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/categories?post=944773"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/tags?post=944773"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}