create_model_lstm_cnn_time_dist: Create LSTM/CNN network for combining multiple sequences

In GenomeNet/deepG: Deep Learning for Genome Sequence Data

Create LSTM/CNN network for combining multiple sequences

Description

Creates a network consisting of an arbitrary number of CNN, LSTM and dense layers. Input is a 4D tensor, where axis correspond to:

1. batch size

2. number of samples in one batch

3. length of one sample

4. size of vocabulary

After LSTM/CNN part all representations get aggregated by summation. Can be used to make single prediction for combination of multiple input sequences. Architecture is equivalent to create_model_lstm_cnn_multi_input but instead of multiple input layers with 3D input, input here in one 4D tensor.

Usage

create_model_lstm_cnn_time_dist(
  maxlen = 50,
  dropout_lstm = 0,
  recurrent_dropout_lstm = 0,
  layer_lstm = NULL,
  layer_dense = c(4),
  solver = "adam",
  learning_rate = 0.001,
  vocabulary_size = 4,
  bidirectional = FALSE,
  stateful = FALSE,
  batch_size = NULL,
  compile = TRUE,
  kernel_size = NULL,
  filters = NULL,
  strides = NULL,
  pool_size = NULL,
  padding = "same",
  dilation_rate = NULL,
  gap_time_dist = NULL,
  use_bias = TRUE,
  zero_mask = FALSE,
  label_smoothing = 0,
  label_noise_matrix = NULL,
  last_layer_activation = "softmax",
  loss_fn = "categorical_crossentropy",
  auc_metric = FALSE,
  f1_metric = FALSE,
  samples_per_target,
  batch_norm_momentum = 0.99,
  verbose = TRUE,
  model_seed = NULL,
  aggregation_method = NULL,
  transformer_args = NULL,
  lstm_time_dist = NULL,
  mixed_precision = FALSE,
  bal_acc = FALSE,
  mirrored_strategy = NULL
)

Arguments

maxlen	Length of predictor sequence.
dropout_lstm	Fraction of the units to drop for inputs.
recurrent_dropout_lstm	Fraction of the units to drop for recurrent state.
layer_lstm	Number of cells per network layer. Can be a scalar or vector.
layer_dense	Vector specifying number of neurons per dense layer after last LSTM or CNN layer (if no LSTM used).
solver	Optimization method, options are ⁠"adam", "adagrad", "rmsprop"⁠ or "sgd".
learning_rate	Learning rate for optimizer.
vocabulary_size	Number of unique character in vocabulary.
bidirectional	Use bidirectional wrapper for lstm layers.
stateful	Boolean. Whether to use stateful LSTM layer.
batch_size	Number of samples that are used for one network update. Only used if stateful = TRUE.
compile	Whether to compile the model.
kernel_size	Size of 1d convolutional layers. For multiple layers, assign a vector. (e.g, rep(3,2) for two layers and kernel size 3)
filters	Number of filters. For multiple layers, assign a vector.
strides	Stride values. For multiple layers, assign a vector.
pool_size	Integer, size of the max pooling windows. For multiple layers, assign a vector.
padding	Padding of CNN layers, e.g. ⁠"same", "valid"⁠ or "causal".
dilation_rate	Integer, the dilation rate to use for dilated convolution.
gap_time_dist	Pooling or flatten method after last time distribution wrapper. Same options as for flatten_method argument in create_model_transformer function.
use_bias	Boolean. Usage of bias for CNN layers.
zero_mask	Boolean, whether to apply zero masking before LSTM layer. Only used if model does not use any CNN layers.
label_smoothing	Float in [0, 1]. If 0, no smoothing is applied. If > 0, loss between the predicted labels and a smoothed version of the true labels, where the smoothing squeezes the labels towards 0.5. The closer the argument is to 1 the more the labels get smoothed.
label_noise_matrix	Matrix of label noises. Every row stands for one class and columns for percentage of labels in that class. If first label contains 5 percent wrong labels and second label no noise, then label_noise_matrix <- matrix(c(0.95, 0.05, 0, 1), nrow = 2, byrow = TRUE )
last_layer_activation	Activation function of output layer(s). For example "sigmoid" or "softmax".
loss_fn	Either "categorical_crossentropy" or "binary_crossentropy". If label_noise_matrix given, will use custom "noisy_loss".
auc_metric	Whether to add AUC metric.
f1_metric	Whether to add F1 metric.
samples_per_target	Number of samples to combine for one target.
batch_norm_momentum	Momentum for the moving mean and the moving variance.
verbose	Boolean.
model_seed	Set seed for model parameters in tensorflow if not NULL.
aggregation_method	At least one of the options ⁠"sum", "mean", "max"⁠.
transformer_args	List of arguments for transformer blocks; see layer_transformer_block_wrapper. Additionally, list can contain pool_flatten argument to apply global pooling or flattening after last transformer block (same options as flatten_method argument in create_model_transformer function).
lstm_time_dist	Vector containing number of units per LSTM cell. Applied after time distribution part.
mixed_precision	Whether to use mixed precision (https://www.tensorflow.org/guide/mixed_precision).
bal_acc	Whether to add balanced accuracy.
mirrored_strategy	Whether to use distributed mirrored strategy. If NULL, will use distributed mirrored strategy only if >1 GPU available.

Value

A keras model with time distribution wrapper applied to LSTM and CNN layers.

Examples

# Examples needs keras attached to run 
maxlen <- 50

library(keras)
create_model_lstm_cnn_time_dist(
  maxlen = maxlen,
  vocabulary_size = 4,
  samples_per_target = 7,
  kernel_size = c(10, 10),
  filters = c(64, 128),
  pool_size = c(2, 2),
  layer_lstm = c(32),
  aggregation_method = c("max"),
  layer_dense = c(64, 2),
  learning_rate = 0.001)
 
 

GenomeNet/deepG documentation built on Dec. 24, 2024, 12:11 p.m.