Man pages for rstudio/keras
R Interface to 'Keras'

activation_reluActivation functions
application_inception_resnet_v2Inception-ResNet v2 model, with weights trained on ImageNet
application_inception_v3Inception V3 model, with weights pre-trained on ImageNet.
application_mobilenetMobileNet model architecture.
application_resnet50ResNet50 model for Keras.
application_vggVGG16 and VGG19 models for Keras.
application_xceptionXception V1 model for Keras.
backendKeras backend tensor engine
bidirectionalBidirectional wrapper for RNNs.
callback_csv_loggerCallback that streams epoch results to a csv file
callback_early_stoppingStop training when a monitored quantity has stopped...
callback_lambdaCreate a custom callback
callback_learning_rate_schedulerLearning rate scheduler.
callback_model_checkpointSave the model after every epoch.
callback_progbar_loggerCallback that prints metrics to stdout.
callback_reduce_lr_on_plateauReduce learning rate when a metric has stopped improving.
callback_remote_monitorCallback used to stream events to a server.
callback_tensorboardTensorBoard basic visualizations
callback_terminate_on_naanCallback that terminates training when a NaN loss is...
clone_modelClone a model instance.
compileConfigure a Keras model for training
constraint_maxnormMaxNorm weight constraint
constraint_minmaxnormMinMaxNorm weight constraint
constraint_nonnegNonNeg weight constraint
constraint_unitnormUnitNorm weight constraint
count_paramsCount the total number of scalars composing the weights.
create_layerCreate a Keras Layer
dataset_boston_housingBoston housing price regression dataset
dataset_cifar10CIFAR10 small image classification
dataset_cifar100CIFAR100 small image classification
dataset_fashion_mnistFashion-MNIST database of fashion articles
dataset_imdbIMDB Movie reviews sentiment classification
dataset_mnistMNIST database of handwritten digits
dataset_reutersReuters newswire topics classification
evaluate_generatorEvaluates the model on a data generator.
evaluate.keras.engine.training.ModelEvaluate a Keras model
export_savedmodel.keras.engine.training.ModelExport a Saved Model
fitTrain a Keras model
fit_generatorFits the model on data yielded batch-by-batch by a generator.
fit_image_data_generatorFit image data generator internal statistics to some sample...
fit_text_tokenizerUpdate tokenizer internal vocabulary based on a list of texts...
flow_images_from_dataGenerates batches of augmented/normalized data from image...
flow_images_from_directoryGenerates batches of data from images in a directory (with...
freeze_weightsFreeze and unfreeze weights
generator_nextRetreive the next item from a generator
get_configLayer/Model configuration
get_fileDownloads a file from a URL if it not already in the cache.
get_input_atRetrieve tensors for layers with multiple nodes
get_layerRetrieves a layer based on either its name (unique) or index.
get_weightsLayer/Model weights as R arrays
hdf5_matrixRepresentation of HDF5 dataset to be used instead of an R...
image_data_generatorGenerate minibatches of image data with real-time data...
image_loadLoads an image into PIL format.
imagenet_decode_predictionsDecodes the prediction of an ImageNet model.
imagenet_preprocess_inputPreprocesses a tensor encoding a batch of images.
image_to_array3D array representation of images
implementationKeras implementation
initializer_constantInitializer that generates tensors initialized to a constant...
initializer_glorot_normalGlorot normal initializer, also called Xavier normal...
initializer_glorot_uniformGlorot uniform initializer, also called Xavier uniform...
initializer_he_normalHe normal initializer.
initializer_he_uniformHe uniform variance scaling initializer.
initializer_identityInitializer that generates the identity matrix.
initializer_lecun_normalLeCun normal initializer.
initializer_lecun_uniformLeCun uniform initializer.
initializer_onesInitializer that generates tensors initialized to 1.
initializer_orthogonalInitializer that generates a random orthogonal matrix.
initializer_random_normalInitializer that generates tensors with a normal...
initializer_random_uniformInitializer that generates tensors with a uniform...
initializer_truncated_normalInitializer that generates a truncated normal distribution.
initializer_variance_scalingInitializer capable of adapting its scale to the shape of...
initializer_zerosInitializer that generates tensors initialized to 0.
install_kerasInstall Keras and the TensorFlow backend
is_keras_availableCheck if Keras is Available
k_absElement-wise absolute value.
k_allBitwise reduction (logical AND).
k_anyBitwise reduction (logical OR).
k_arangeCreates a 1D tensor containing a sequence of integers.
k_argmaxReturns the index of the maximum value along an axis.
k_argminReturns the index of the minimum value along an axis.
k_backendActive Keras backend
k_batch_dotBatchwise dot product.
k_batch_flattenTurn a nD tensor into a 2D tensor with same 1st dimension.
k_batch_get_valueReturns the value of more than one tensor variable.
k_batch_normalizationApplies batch normalization on x given mean, var, beta and...
k_batch_set_valueSets the values of many tensor variables at once.
k_bias_addAdds a bias vector to a tensor.
k_binary_crossentropyBinary crossentropy between an output tensor and a target...
k_castCasts a tensor to a different dtype and returns it.
k_cast_to_floatxCast an array to the default Keras float type.
k_categorical_crossentropyCategorical crossentropy between an output tensor and a...
k_clear_sessionDestroys the current TF graph and creates a new one.
k_clipElement-wise value clipping.
k_concatenateConcatenates a list of tensors alongside the specified axis.
k_constantCreates a constant tensor.
k_conv1d1D convolution.
k_conv2d2D convolution.
k_conv2d_transpose2D deconvolution (i.e. transposed convolution).
k_conv3d3D convolution.
k_conv3d_transpose3D deconvolution (i.e. transposed convolution).
k_cosComputes cos of x element-wise.
k_count_paramsReturns the static number of elements in a Keras variable or...
k_ctc_batch_costRuns CTC loss algorithm on each batch element.
k_ctc_decodeDecodes the output of a softmax.
k_ctc_label_dense_to_sparseConverts CTC labels from dense to sparse.
k_cumprodCumulative product of the values in a tensor, alongside the...
k_cumsumCumulative sum of the values in a tensor, alongside the...
k_depthwise_conv2dDepthwise 2D convolution with separable filters.
k_dotMultiplies 2 tensors (and/or variables) and returns a...
k_dropoutSets entries in 'x' to zero at random, while scaling the...
k_dtypeReturns the dtype of a Keras tensor or variable, as a string.
k_eluExponential linear unit.
k_epsilonFuzz factor used in numeric expressions.
k_equalElement-wise equality between two tensors.
keras_arrayKeras array object
KerasCallbackBase R6 class for Keras callbacks
KerasLayerBase R6 class for Keras layers
keras_modelKeras Model
keras_model_sequentialKeras Model composed of a linear stack of layers
keras-packageR interface to Keras
k_evalEvaluates the value of a variable.
k_expElement-wise exponential.
k_expand_dimsAdds a 1-sized dimension at index 'axis'.
k_eyeInstantiate an identity matrix and returns it.
k_flattenFlatten a tensor.
k_floatxDefault float type
k_foldlReduce elems using fn to combine them from left to right.
k_foldrReduce elems using fn to combine them from right to left.
k_functionInstantiates a Keras function
k_gatherRetrieves the elements of indices 'indices' in the tensor...
k_get_sessionTF session to be used by the backend.
k_get_uidGet the uid for the default graph.
k_get_valueReturns the value of a variable.
k_get_variable_shapeReturns the shape of a variable.
k_gradientsReturns the gradients of 'variables' w.r.t. 'loss'.
k_greaterElement-wise truth value of (x > y).
k_greater_equalElement-wise truth value of (x >= y).
k_hard_sigmoidSegment-wise linear approximation of sigmoid.
k_identityReturns a tensor with the same content as the input tensor.
k_image_data_formatDefault image data format convention ('channels_first' or...
k_in_test_phaseSelects 'x' in test phase, and 'alt' otherwise.
k_in_top_kReturns whether the 'targets' are in the top 'k'...
k_in_train_phaseSelects 'x' in train phase, and 'alt' otherwise.
k_int_shapeReturns the shape of tensor or variable as a list of int or...
k_is_keras_tensorReturns whether 'x' is a Keras tensor.
k_is_placeholderReturns whether 'x' is a placeholder.
k_is_sparseReturns whether a tensor is a sparse tensor.
k_l2_normalizeNormalizes a tensor wrt the L2 norm alongside the specified...
k_learning_phaseReturns the learning phase flag.
k_lessElement-wise truth value of (x < y).
k_less_equalElement-wise truth value of (x <= y).
k_local_conv1dApply 1D conv with un-shared weights.
k_local_conv2dApply 2D conv with un-shared weights.
k_logElement-wise log.
k_logsumexpComputes log(sum(exp(elements across dimensions of a...
k_manual_variable_initializationSets the manual variable initialization flag.
k_map_fnMap the function fn over the elements elems and return the...
k_maxMaximum value in a tensor.
k_maximumElement-wise maximum of two tensors.
k_meanMean of a tensor, alongside the specified axis.
k_minMinimum value in a tensor.
k_minimumElement-wise minimum of two tensors.
k_moving_average_updateCompute the moving average of a variable.
k_ndimReturns the number of axes in a tensor, as an integer.
k_normalize_batch_in_trainingComputes mean and std for batch then apply...
k_not_equalElement-wise inequality between two tensors.
k_one_hotComputes the one-hot representation of an integer tensor.
k_onesInstantiates an all-ones tensor variable and returns it.
k_ones_likeInstantiates an all-ones variable of the same shape as...
k_permute_dimensionsPermutes axes in a tensor.
k_placeholderInstantiates a placeholder tensor and returns it.
k_pool2d2D Pooling.
k_pool3d3D Pooling.
k_powElement-wise exponentiation.
k_print_tensorPrints 'message' and the tensor value when evaluated.
k_prodMultiplies the values in a tensor, alongside the specified...
k_random_binomialReturns a tensor with random binomial distribution of values.
k_random_normalReturns a tensor with normal distribution of values.
k_random_normal_variableInstantiates a variable with values drawn from a normal...
k_random_uniformReturns a tensor with uniform distribution of values.
k_random_uniform_variableInstantiates a variable with values drawn from a uniform...
k_reluRectified linear unit.
k_repeatRepeats a 2D tensor.
k_repeat_elementsRepeats the elements of a tensor along an axis.
k_reset_uidsReset graph identifiers.
k_reshapeReshapes a tensor to the specified shape.
k_resize_imagesResizes the images contained in a 4D tensor.
k_resize_volumesResizes the volume contained in a 5D tensor.
k_reverseReverse a tensor along the specified axes.
k_rnnIterates over the time dimension of a tensor
k_roundElement-wise rounding to the closest integer.
k_separable_conv2d2D convolution with separable filters.
k_set_learning_phaseSets the learning phase to a fixed value.
k_set_valueSets the value of a variable, from an R array.
k_shapeReturns the symbolic shape of a tensor or variable.
k_sigmoidElement-wise sigmoid.
k_signElement-wise sign.
k_sinComputes sin of x element-wise.
k_softmaxSoftmax of a tensor.
k_softplusSoftplus of a tensor.
k_softsignSoftsign of a tensor.
k_sparse_categorical_crossentropyCategorical crossentropy with integer targets.
k_spatial_2d_paddingPads the 2nd and 3rd dimensions of a 4D tensor.
k_spatial_3d_paddingPads 5D tensor with zeros along the depth, height, width...
k_sqrtElement-wise square root.
k_squareElement-wise square.
k_squeezeRemoves a 1-dimension from the tensor at index 'axis'.
k_stackStacks a list of rank 'R' tensors into a rank 'R+1' tensor.
k_stdStandard deviation of a tensor, alongside the specified axis.
k_stop_gradientReturns 'variables' but with zero gradient w.r.t. every other...
k_sumSum of the values in a tensor, alongside the specified axis.
k_switchSwitches between two operations depending on a scalar value.
k_tanhElement-wise tanh.
k_temporal_paddingPads the middle dimension of a 3D tensor.
k_tileCreates a tensor by tiling 'x' by 'n'.
k_to_denseConverts a sparse tensor into a dense tensor and returns it.
k_transposeTransposes a tensor and returns it.
k_truncated_normalReturns a tensor with truncated random normal distribution of...
k_updateUpdate the value of 'x' to 'new_x'.
k_update_addUpdate the value of 'x' by adding 'increment'.
k_update_subUpdate the value of 'x' by subtracting 'decrement'.
k_varVariance of a tensor, alongside the specified axis.
k_variableInstantiates a variable and returns it.
k_zerosInstantiates an all-zeros variable and returns it.
k_zeros_likeInstantiates an all-zeros variable of the same shape as...
layer_activationApply an activation function to an output.
layer_activation_eluExponential Linear Unit.
layer_activation_leaky_reluLeaky version of a Rectified Linear Unit.
layer_activation_parametric_reluParametric Rectified Linear Unit.
layer_activation_thresholded_reluThresholded Rectified Linear Unit.
layer_activity_regularizationLayer that applies an update to the cost function based input...
layer_addLayer that adds a list of inputs.
layer_alpha_dropoutApplies Alpha Dropout to the input.
layer_averageLayer that averages a list of inputs.
layer_average_pooling_1dAverage pooling for temporal data.
layer_average_pooling_2dAverage pooling operation for spatial data.
layer_average_pooling_3dAverage pooling operation for 3D data (spatial or...
layer_batch_normalizationBatch normalization layer (Ioffe and Szegedy, 2014).
layer_concatenateLayer that concatenates a list of inputs.
layer_conv_1d1D convolution layer (e.g. temporal convolution).
layer_conv_2d2D convolution layer (e.g. spatial convolution over images).
layer_conv_2d_transposeTransposed 2D convolution layer (sometimes called...
layer_conv_3d3D convolution layer (e.g. spatial convolution over volumes).
layer_conv_3d_transposeTransposed 3D convolution layer (sometimes called...
layer_conv_lstm_2dConvolutional LSTM.
layer_cropping_1dCropping layer for 1D input (e.g. temporal sequence).
layer_cropping_2dCropping layer for 2D input (e.g. picture).
layer_cropping_3dCropping layer for 3D data (e.g. spatial or spatio-temporal).
layer_cudnn_gruFast GRU implementation backed by CuDNN.
layer_cudnn_lstmFast LSTM implementation backed by CuDNN.
layer_denseAdd a densely-connected NN layer to an output
layer_dotLayer that computes a dot product between samples in two...
layer_dropoutApplies Dropout to the input.
layer_embeddingTurns positive integers (indexes) into dense vectors of fixed...
layer_flattenFlattens an input
layer_gaussian_dropoutApply multiplicative 1-centered Gaussian noise.
layer_gaussian_noiseApply additive zero-centered Gaussian noise.
layer_global_average_pooling_1dGlobal average pooling operation for temporal data.
layer_global_average_pooling_2dGlobal average pooling operation for spatial data.
layer_global_average_pooling_3dGlobal Average pooling operation for 3D data.
layer_global_max_pooling_1dGlobal max pooling operation for temporal data.
layer_global_max_pooling_2dGlobal max pooling operation for spatial data.
layer_global_max_pooling_3dGlobal Max pooling operation for 3D data.
layer_gruGated Recurrent Unit - Cho et al.
layer_inputInput layer
layer_lambdaWraps arbitrary expression as a layer
layer_locally_connected_1dLocally-connected layer for 1D inputs.
layer_locally_connected_2dLocally-connected layer for 2D inputs.
layer_lstmLong-Short Term Memory unit - Hochreiter 1997.
layer_maskingMasks a sequence by using a mask value to skip timesteps.
layer_maximumLayer that computes the maximum (element-wise) a list of...
layer_max_pooling_1dMax pooling operation for temporal data.
layer_max_pooling_2dMax pooling operation for spatial data.
layer_max_pooling_3dMax pooling operation for 3D data (spatial or...
layer_minimumLayer that computes the minimum (element-wise) a list of...
layer_multiplyLayer that multiplies (element-wise) a list of inputs.
layer_permutePermute the dimensions of an input according to a given...
layer_repeat_vectorRepeats the input n times.
layer_reshapeReshapes an output to a certain shape.
layer_separable_conv_2dDepthwise separable 2D convolution.
layer_simple_rnnFully-connected RNN where the output is to be fed back to...
layer_spatial_dropout_1dSpatial 1D version of Dropout.
layer_spatial_dropout_2dSpatial 2D version of Dropout.
layer_spatial_dropout_3dSpatial 3D version of Dropout.
layer_subtractLayer that subtracts two inputs.
layer_upsampling_1dUpsampling layer for 1D inputs.
layer_upsampling_2dUpsampling layer for 2D inputs.
layer_upsampling_3dUpsampling layer for 3D inputs.
layer_zero_padding_1dZero-padding layer for 1D input (e.g. temporal sequence).
layer_zero_padding_2dZero-padding layer for 2D input (e.g. picture).
layer_zero_padding_3dZero-padding layer for 3D data (spatial or spatio-temporal).
loss_mean_squared_errorModel loss functions
make_sampling_tableGenerates a word rank-based probabilistic sampling table.
metric_binary_accuracyModel performance metrics
model_to_jsonModel configuration as JSON
model_to_yamlModel configuration as YAML
multi-assignAssign values to names
multi_gpu_modelReplicates a model on different GPUs.
normalizeNormalize a matrix or nd-array
optimizer_adadeltaAdadelta optimizer.
optimizer_adagradAdagrad optimizer.
optimizer_adamAdam optimizer
optimizer_adamaxAdamax optimizer
optimizer_nadamNesterov Adam optimizer
optimizer_rmspropRMSProp optimizer
optimizer_sgdStochastic gradient descent optimizer
pad_sequencesPads each sequence to the same length (length of the longest...
pipePipe operator
plot.keras_training_historyPlot training history
pop_layerRemove the last layer in a model
predict_generatorGenerates predictions for the input samples from a data...
predict.keras.engine.training.ModelGenerate predictions from a Keras model
predict_on_batchReturns predictions for a single batch of samples.
predict_probaGenerates probability or class probability predictions for...
reexportsObjects exported from other packages
regularizer_l1L1 and L2 regularization
reset_statesReset the states for a layer
save_model_hdf5Save/Load models using HDF5 files
save_model_weights_hdf5Save/Load model weights using HDF5 files
save_text_tokenizerSave a text tokenizer to an external file
sequences_to_matrixConvert a list of sequences into a matrix.
serialize_modelSerialize a model to an R object
skipgramsGenerates skipgram word pairs.
summary.keras.engine.training.ModelPrint a summary of a Keras model
text_hashing_trickConverts a text to a sequence of indexes in a fixed-size...
text_one_hotOne-hot encode a text into a list of word indexes in a...
texts_to_matrixConvert a list of texts to a matrix.
texts_to_sequencesTransform each text in texts in a sequence of integers.
texts_to_sequences_generatorTransforms each text in texts in a sequence of integers.
text_tokenizerText tokenization utility
text_to_word_sequenceConvert text to a sequence of words (or tokens).
time_distributedApply a layer to every temporal slice of an input.
to_categoricalConverts a class vector (integers) to binary class matrix.
train_on_batchSingle gradient update or model evaluation over one batch of...
use_implementationSelect a Keras implementation and backend
with_custom_object_scopeProvide a scope with mappings of names to custom objects
rstudio/keras documentation built on Dec. 15, 2017, 2:12 p.m.