diff options
author | 2016-10-10 10:26:22 -0800 | |
---|---|---|
committer | 2016-10-10 11:35:00 -0700 | |
commit | edaf3b342db4afa1c872da541fb0ac176a4e8ef9 (patch) | |
tree | b976073fdc2a6404cbdc3ee323a637e2e1b16846 /tensorflow/g3doc/tutorials/deep_cnn/index.md | |
parent | d1518c26530daaee854e73365bd7dfb9a2f69dbd (diff) |
Merge changes from github.
Change: 135698415
Diffstat (limited to 'tensorflow/g3doc/tutorials/deep_cnn/index.md')
-rw-r--r-- | tensorflow/g3doc/tutorials/deep_cnn/index.md | 34 |
1 files changed, 16 insertions, 18 deletions
diff --git a/tensorflow/g3doc/tutorials/deep_cnn/index.md b/tensorflow/g3doc/tutorials/deep_cnn/index.md index d1ef5f6405..89ba53ac6f 100644 --- a/tensorflow/g3doc/tutorials/deep_cnn/index.md +++ b/tensorflow/g3doc/tutorials/deep_cnn/index.md @@ -32,17 +32,15 @@ new ideas and experimenting with new techniques. The CIFAR-10 tutorial demonstrates several important constructs for designing larger and more sophisticated models in TensorFlow: -* Core mathematical components including [convolution]( -../../api_docs/python/nn.md#conv2d) ([wiki]( -https://en.wikipedia.org/wiki/Convolution)), [rectified linear activations]( -../../api_docs/python/nn.md#relu) ([wiki]( -https://en.wikipedia.org/wiki/Rectifier_(neural_networks))), [max pooling]( -../../api_docs/python/nn.md#max_pool) ([wiki]( -https://en.wikipedia.org/wiki/Convolutional_neural_network#Pooling_layer)) -and [local response normalization]( -../../api_docs/python/nn.md#local_response_normalization) -(Chapter 3.3 in [AlexNet paper]( -http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf)). +* Core mathematical components including [convolution](../../api_docs/python/nn.md#conv2d) +([wiki](https://en.wikipedia.org/wiki/Convolution)), +[rectified linear activations](../../api_docs/python/nn.md#relu) +([wiki](https://en.wikipedia.org/wiki/Rectifier_(neural_networks))), +[max pooling](../../api_docs/python/nn.md#max_pool) +([wiki](https://en.wikipedia.org/wiki/Convolutional_neural_network#Pooling_layer)) +and [local response normalization](../../api_docs/python/nn.md#local_response_normalization) +(Chapter 3.3 in +[AlexNet paper](http://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf)). * [Visualization](../../how_tos/summaries_and_tensorboard/index.md) of network activities during training, including input images, losses and distributions of activations and gradients. @@ -57,7 +55,7 @@ that systematically decrements over time. for input data to isolate the model from disk latency and expensive image pre-processing. -We also provide a [multi-GPU version](#training-a-model-using-multiple-gpu-cards) +We also provide a [multi-GPU version](#training-a-model-using-multiple-gpu-cards) of the model which demonstrates: * Configuring a model to train across multiple GPU cards in parallel. @@ -111,7 +109,7 @@ adds operations that perform inference, i.e. classification, on supplied images. add operations that compute the loss, gradients, variable updates and visualization summaries. -### Model Inputs +### Model Inputs The input part of the model is built by the functions `inputs()` and `distorted_inputs()` which read images from the CIFAR-10 binary data files. @@ -149,7 +147,7 @@ processing time. To prevent these operations from slowing down training, we run them inside 16 separate threads which continuously fill a TensorFlow [queue](../../api_docs/python/io_ops.md#shuffle_batch). -### Model Prediction +### Model Prediction The prediction part of the model is constructed by the `inference()` function which adds operations to compute the *logits* of the predictions. That part of @@ -174,8 +172,8 @@ Here is a graph generated from TensorBoard describing the inference operation: </div> > **EXERCISE**: The output of `inference` are un-normalized logits. Try editing -the network architecture to return normalized predictions using [`tf.nn.softmax()`] -(../../api_docs/python/nn.md#softmax). +the network architecture to return normalized predictions using +[`tf.nn.softmax()`](../../api_docs/python/nn.md#softmax). The `inputs()` and `inference()` functions provide all the components necessary to perform evaluation on a model. We now shift our focus towards @@ -188,7 +186,7 @@ layers of Alex's original model are locally connected and not fully connected. Try editing the architecture to exactly reproduce the locally connected architecture in the top layer. -### Model Training +### Model Training The usual method for training a network to perform N-way classification is [multinomial logistic regression](https://en.wikipedia.org/wiki/Multinomial_logistic_regression), @@ -307,7 +305,7 @@ values. See how the scripts use [`ExponentialMovingAverage`](../../api_docs/python/train.md#ExponentialMovingAverage) for this purpose. -## Evaluating a Model +## Evaluating a Model Let us now evaluate how well the trained model performs on a hold-out data set. The model is evaluated by the script `cifar10_eval.py`. It constructs the model |