ProcData (version 0.2.5)

seq2feature_seq2seq: Feature Extraction by autoencoder

Description

seq2feature_seq2seq extract features from response processes by autoencoder.

Usage

seq2feature_seq2seq(seqs, ae_type = "action", K, rnn_type = "lstm",
  n_epoch = 50, method = "last", step_size = 1e-04,
  optimizer_name = "adam", cumulative = FALSE, log = TRUE,
  weights = c(1, 0.5), samples_train, samples_valid,
  samples_test = NULL, pca = TRUE, gpu = FALSE, parallel = FALSE,
  seed = 12345, verbose = TRUE, return_theta = TRUE)

Arguments

seqs

an object of class "proc".

ae_type

a string specifies the type of autoencoder. The autoencoder can be an action sequence autoencoder ("action"), a time sequence autoencoder ("time"), or an action-time sequence autoencoder ("both").

K

the number of features to be extracted.

rnn_type

the type of recurrent unit to be used for modeling response processes. "lstm" for the long-short term memory unit. "gru" for the gated recurrent unit.

n_epoch

the number of training epochs for the autoencoder.

method

the method for computing features from the output of an recurrent neural network in the encoder. Available options are "last" and "avg".

step_size

the learning rate of optimizer.

optimizer_name

a character string specifying the optimizer to be used for training. Availabel options are "sgd", "rmsprop", "adadelta", and "adam".

cumulative

logical. If TRUE, the sequence of cumulative time up to each event is used as input to the neural network. If FALSE, the sequence of inter-arrival time (gap time between an event and the previous event) will be used as input to the neural network. Default is FALSE.

log

logical. If TRUE, for the timestamp sequences, input of the neural net is the base-10 log of the original sequence of times plus 1 (i.e., log10(t+1)). If FALSE, the original sequence of times is used.

weights

a vector of 2 elements for the weight of the loss of action sequences (categorical_crossentropy) and time sequences (mean squared error), respectively. The total loss is calculated as the weighted sum of the two losses.

samples_train, samples_valid, samples_test

vectors of indices specifying the training, validation and test sets for training autoencoder.

pca

logical. If TRUE, the principal components of features are returned. Default is TRUE.

gpu

logical. If TRUE, use gpu for training when available. Default is FALSE.

parallel

logical. If TRUE, allow cpu parallel computing. Default is FALSE.

seed

random seed.

verbose

logical. If TRUE, training progress is printed.

return_theta

logical. If TRUE, extracted features are returned.

Value

seq2feature_seq2seq returns a list containing

theta

a matrix containing K features or principal features. Each column is a feature.

train_loss

a vector of length n_epoch recording the trace of training losses.

valid_loss

a vector of length n_epoch recording the trace of validation losses.

test_loss

a vector of length n_epoch recording the trace of test losses. Exists only if samples_test is not NULL.

Details

This function wraps aseq2feature_seq2seq, tseq2feature_seq2seq, and atseq2feature_seq2seq.

See Also

chooseK_seq2seq for choosing K through cross-validation.

Other feature extraction methods: aseq2feature_seq2seq, atseq2feature_seq2seq, seq2feature_mds_large, seq2feature_mds, tseq2feature_seq2seq

Examples

# NOT RUN {
n <- 50
data(cc_data)
samples <- sample(1:length(cc_data$seqs$time_seqs), n)
seqs <- sub_seqs(cc_data$seqs, samples)

# action sequence autoencoder
K_res <- chooseK_seq2seq(seqs=seqs, ae_type="action", K_cand=c(5, 10), 
                         n_epoch=5, n_fold=2, valid_prop=0.2)
seq2seq_res <- seq2feature_seq2seq(seqs=seqs, ae_type="action", K=K_res$K, 
                       n_epoch=5, samples_train=1:40, samples_valid=41:50)
theta <- seq2seq_res$theta

# time sequence autoencoder
K_res <- chooseK_seq2seq(seqs=seqs, ae_type="time", K_cand=c(5, 10), 
                         n_epoch=5, n_fold=2, valid_prop=0.2)
seq2seq_res <- seq2feature_seq2seq(seqs=seqs, ae_type="time", K=K_res$K, 
                       n_epoch=5, samples_train=1:40, samples_valid=41:50)
theta <- seq2seq_res$theta

# action and time sequence autoencoder
K_res <- chooseK_seq2seq(seqs=seqs, ae_type="both", K_cand=c(5, 10), 
                         n_epoch=5, n_fold=2, valid_prop=0.2)
seq2seq_res <- seq2feature_seq2seq(seqs=seqs, ae_type="both", K=K_res$K, 
                       n_epoch=5, samples_train=1:40, samples_valid=41:50)
theta <- seq2seq_res$theta
plot(seq2seq_res$train_loss, col="blue", type="l")
lines(seq2seq_res$valid_loss, col="red")
# }