Improve denoising AE performance

R/autoencoder.R

@@ -222,24 +222,40 @@ train.ruta_autoencoder <- function(
     metrics = metrics
   )
 
-  input_data <- if (is.null(learner$filter)) {
-    data
-  } else {
-    apply_filter(learner$filter, data)
-  }
+  # input_data <- if (is.null(learner$filter)) {
+  #   data
+  # } else {
+  #   apply_filter(learner$filter, data)
+  # }
 
   if (!is.null(validation_data)) {
     validation_data <- list(validation_data, validation_data)
   }
 
-  keras::fit(
-    learner$models$autoencoder,
-    x = input_data,
-    y = data,
-    epochs = epochs,
-    validation_data = validation_data,
-    ...
-  )
+  if (is.null(learner$filter)) {
+    keras::fit(
+      learner$models$autoencoder,
+      x = data,
+      y = data,
+      epochs = epochs,
+      validation_data = validation_data,
+      ...
+    )
+  } else {
+    batch_size <- list(...)$batch_size
+    if (is.null(batch_size)) batch_size <- 32
+
+    keras::fit_generator(
+      learner$models$autoencoder,
+      to_keras(learner$filter, data, batch_size),
+      steps_per_epoch = ceiling(dim(data)[1] / batch_size),
+      epochs = epochs,
+      validation_data = validation_data,
+      ...
+    )
+  }
+
+
 
   invisible(learner)
 }

R/filter.R

@@ -161,3 +161,25 @@ apply_filter.ruta_noise_cauchy <- function(filter, data, ...) {
 
   data + term
 }
+
+#' @import R.utils
+#' @param data
+#' @param batch_size
+to_keras.ruta_filter <- function(x, data, batch_size, ...) {
+  limit <- dim(data)[1]
+  order <- sample.int(limit)
+  start <- 1
+  function() {
+    if (start + batch_size > limit) {
+      idx <- order[start:limit]
+      order <- sample.int(limit)
+      start <- 1
+    } else {
+      idx <- order[start:(start + batch_size - 1)]
+      start <- start + batch_size
+    }
+    original <- R.utils::extract(data, "1" = idx)
+    noisy <- apply_filter(x, original)
+    list(noisy, original)
+  }
+}