Support All keras activations functions (#1244)

* allow all keras activations

* update news

* add tests

---------

Co-authored-by: ‘topepo’ <[email protected]>

Author: EmilHvitfeldt

NAMESPACE

@@ -253,6 +253,7 @@ export(glm_grouped)
 export(has_multi_predict)
 export(importance_weights)
 export(is_varying)
+export(keras_activations)
 export(keras_mlp)
 export(keras_predict_classes)
 export(knit_engine_docs)

NEWS.md

@@ -15,6 +15,8 @@
 
 * New `extract_fit_time()` method has been added that returns the time it took to train the model (#853).
 
+* `mlp()` with `keras` engine now work for all activation functions currently supported by `keras` (#1127).
+
 ## Other Changes
 
 * Transitioned package errors and warnings to use cli (#1147 and #1148 by @shum461, #1153 by @RobLBaker and @wright13, #1154 by @JamesHWade, #1160, #1161, #1081).

R/mlp.R

@@ -192,8 +192,13 @@ keras_mlp <-
            seeds = sample.int(10^5, size = 3),
            ...) {
 
-    act_funs <- c("linear", "softmax", "relu", "elu", "tanh")
-    rlang::arg_match(activation, act_funs)
+    allowed_keras_activation <- keras_activations()
+    good_activation <- activation %in% allowed_keras_activation
+    if (!all(good_activation)) {
+      cli::cli_abort(
+        "{.arg activation} should be one of: {allowed_activation}."
+      )
+    }
 
     if (penalty > 0 & dropout > 0) {
       cli::cli_abort("Please use either dropout or weight decay.", call = NULL)
@@ -344,6 +349,19 @@ mlp_num_weights <- function(p, hidden_units, classes) {
   ((p + 1) * hidden_units) + ((hidden_units+1) * classes)
 }
 
+allowed_keras_activation <-
+ c("elu", "exponential", "gelu", "hard_sigmoid", "linear", "relu", "selu", 
+   "sigmoid", "softmax", "softplus", "softsign", "swish", "tanh")
+
+#' Activation functions for neural networks in keras
+#'
+#' @keywords internal
+#' @return A character vector of values.
+#' @export
+keras_activations <- function() {
+  allowed_keras_activation
+}
+
 ## -----------------------------------------------------------------------------
 
 #' @importFrom purrr map

man/keras_activations.Rd

@@ -149,11 +149,10 @@ car_basic <- mlp(mode = "regression", epochs = 10) %>%
 
 bad_keras_reg <-
   mlp(mode = "regression") %>%
-  set_engine("keras", min.node.size = -10)
+  set_engine("keras", min.node.size = -10, verbose = 0)
 
 # ------------------------------------------------------------------------------
 
-
 test_that('keras execution, regression', {
   skip_on_cran()
   skip_if_not_installed("keras")
@@ -211,7 +210,6 @@ test_that('keras regression prediction', {
   keras::backend()$clear_session()
 })
 
-
 # ------------------------------------------------------------------------------
 
 test_that('multivariate nnet formula', {
@@ -247,3 +245,34 @@ test_that('multivariate nnet formula', {
 
   keras::backend()$clear_session()
 })
+
+# ------------------------------------------------------------------------------
+
+test_that('all keras activation functions', {
+  skip_on_cran()
+  skip_if_not_installed("keras")
+  skip_if_not_installed("modeldata")
+  skip_if(!is_tf_ok())
+
+  act <- parsnip:::keras_activations()
+
+  test_act <- function(fn) {
+    set.seed(1)
+    try(
+      mlp(mode = "classification", hidden_units = 2, penalty = 0.01, epochs = 2,
+          activation = !!fn)  %>%
+        set_engine("keras", verbose = 0) %>%
+        parsnip::fit(Class ~ A + B, data = modeldata::two_class_dat),
+      silent = TRUE)
+
+  }
+  test_act_sshhh <- purrr::quietly(test_act)
+
+  for (i in act) {
+    keras::backend()$clear_session()
+    act_res <- test_act_sshhh(i)
+    expect_s3_class(act_res$result, "model_fit")
+    keras::backend()$clear_session()
+  }
+
+})