安装tensorflow:install_tensorflow()
此时会提示安装miniconda3. 而由于miniconda服务器问题,在R中下载miniconda总是超时无法连接。
这时要直接从官网下载miniconda3.
下载exe文件安装以后,还需要对miniconda进行配置,否则依然无法安装tensorflow。
从开始中打开anaconda prompt(miniconda3),进行配置:
conda create -n tensorflow2.6 ###当前tensorflow最高版本为2.6,自己安装时根据实际情况选择版本
然后会自动安装tensorflow环境:
Collecting package metadata (current_repodata.json): doneSolving environment: done ## Package Plan ## environment location: C:\ProgramData\Miniconda3\envs\tensorflow2.6 Proceed ([y]/n)? y Preparing transaction: done Verifying transaction: done Executing transaction: done # # To activate this environment, use # # $ conda activate tensorflow2.6 |
再到R环境下安装tensorflow,就可以成功了。
运行mnist案例:
> network%>%fit(train_images,train_labels,epochs = 5,batch_size = 128)2021-10-10 10:59:22.750466: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2) Epoch 1/5 469/469 [==============================] - 6s 12ms/step - loss: 0.2524 - accuracy: 0.9269 Epoch 2/5 469/469 [==============================] - 6s 12ms/step - loss: 0.1019 - accuracy: 0.9695 Epoch 3/5 469/469 [==============================] - 6s 12ms/step - loss: 0.0679 - accuracy: 0.9797 Epoch 4/5 469/469 [==============================] - 6s 12ms/step - loss: 0.0495 - accuracy: 0.9850 Epoch 5/5 469/469 [==============================] - 6s 12ms/step - loss: 0.0367 - accuracy: 0.9891 > > metrics <- network %>% evaluate(test_images,test_labels) 313/313 [==============================] - 1s 2ms/step - loss: 0.0674 - accuracy: 0.9801 > metrics loss accuracy 0.06739952 0.98009998 > network %>% predict(test_images[1:5,]) %>% k_argmax() tf.Tensor([7 2 1 0 4], shape=(5,), dtype=int64) |
模型精度98%,对测试数据的预测正确。