feat: InputTrafos and OutputTrafos, AcqFunctionEILog (#178)

includes changes to input transformations (#177)

Author: sumny

DESCRIPTION

@@ -81,6 +81,7 @@ Collate:
     'AcqFunctionEHVI.R'
     'AcqFunctionEHVIGH.R'
     'AcqFunctionEI.R'
+    'AcqFunctionEILog.R'
     'AcqFunctionEIPS.R'
     'AcqFunctionMean.R'
     'AcqFunctionMulti.R'
@@ -90,10 +91,17 @@ Collate:
     'AcqFunctionStochasticCB.R'
     'AcqFunctionStochasticEI.R'
     'AcqOptimizer.R'
+    'mlr_input_trafos.R'
+    'InputTrafo.R'
+    'InputTrafoUnitcube.R'
     'aaa.R'
     'OptimizerADBO.R'
     'OptimizerAsyncMbo.R'
     'OptimizerMbo.R'
+    'mlr_output_trafos.R'
+    'OutputTrafo.R'
+    'OutputTrafoLog.R'
+    'OutputTrafoStandardize.R'
     'mlr_result_assigners.R'
     'ResultAssigner.R'
     'ResultAssignerArchive.R'

NAMESPACE

@@ -1,7 +1,9 @@
 # Generated by roxygen2: do not edit by hand
 
 S3method(as.data.table,DictionaryAcqFunction)
+S3method(as.data.table,DictionaryInputTrafo)
 S3method(as.data.table,DictionaryLoopFunction)
+S3method(as.data.table,DictionaryOutputTrafo)
 S3method(as.data.table,DictionaryResultAssigner)
 S3method(print,loop_function)
 export(AcqFunction)
@@ -10,6 +12,7 @@ export(AcqFunctionCB)
 export(AcqFunctionEHVI)
 export(AcqFunctionEHVIGH)
 export(AcqFunctionEI)
+export(AcqFunctionEILog)
 export(AcqFunctionEIPS)
 export(AcqFunctionMean)
 export(AcqFunctionMulti)
@@ -19,9 +22,14 @@ export(AcqFunctionSmsEgo)
 export(AcqFunctionStochasticCB)
 export(AcqFunctionStochasticEI)
 export(AcqOptimizer)
+export(InputTrafo)
+export(InputTrafoUnitcube)
 export(OptimizerADBO)
 export(OptimizerAsyncMbo)
 export(OptimizerMbo)
+export(OutputTrafo)
+export(OutputTrafoLog)
+export(OutputTrafoStandardize)
 export(ResultAssigner)
 export(ResultAssignerArchive)
 export(ResultAssignerSurrogate)
@@ -46,9 +54,13 @@ export(default_loop_function)
 export(default_result_assigner)
 export(default_rf)
 export(default_surrogate)
+export(it)
 export(mlr_acqfunctions)
+export(mlr_input_trafos)
 export(mlr_loop_functions)
+export(mlr_output_trafos)
 export(mlr_result_assigners)
+export(ot)
 export(ras)
 export(redis_available)
 export(srlrn)
@@ -67,6 +79,7 @@ importFrom(stats,pnorm)
 importFrom(stats,quantile)
 importFrom(stats,rexp)
 importFrom(stats,runif)
+importFrom(stats,sd)
 importFrom(stats,setNames)
 importFrom(utils,bibentry)
 useDynLib(mlr3mbo,c_eps_indicator)

R/AcqFunctionAEI.R

@@ -95,8 +95,9 @@ AcqFunctionAEI = R6Class("AcqFunctionAEI",
     #' Update the acquisition function and set `y_effective_best` and `noise_var`.
     update = function() {
       xdt = self$archive$data[, self$archive$cols_x, with = FALSE]
-      p = self$surrogate$predict(xdt)
-      y_effective = p$mean + (self$surrogate_max_to_min * self$constants$values$c * p$se) # pessimistic prediction
+      pred = self$surrogate$predict(xdt)
+      # NOTE: output_trafo_must_be_considered is not relevant to y here because y_effective_best is determined from the predictions
+      y_effective = pred$mean + (self$surrogate_max_to_min * self$constants$values$c * pred$se) # pessimistic prediction
       self$y_effective_best = min(self$surrogate_max_to_min * y_effective)
 
       if (!is.null(self$surrogate$learner$model) && length(self$surrogate$learner$model@covariance@nugget) == 1L) {

R/AcqFunctionEHVI.R

@@ -83,6 +83,9 @@ AcqFunctionEHVI = R6Class("AcqFunctionEHVI",
         stopf("'%s' only works for exactly two objectives.", format(self))
       }
       ys = self$archive$data[, self$archive$cols_y, with = FALSE]
+      if (self$surrogate$output_trafo_must_be_considered) {
+        ys = self$surrogate$output_trafo$transform(ys)
+      }
       for (column in self$archive$cols_y) {
         set(ys, j = column, value = ys[[column]] * self$surrogate_max_to_min[[column]])  # assume minimization
       }
@@ -155,3 +158,4 @@ mlr_acqfunctions$add("ehvi", AcqFunctionEHVI)
 psi_function = function(a, b, mu, sigma) {
   (sigma * dnorm((b - mu) / sigma) + ((a - mu) * pnorm((b - mu) / sigma)))
 }
+

R/AcqFunctionEHVIGH.R

@@ -108,6 +108,9 @@ AcqFunctionEHVIGH = R6Class("AcqFunctionEHVIGH",
     update = function() {
       n_obj = length(self$archive$cols_y)
       ys = self$archive$data[, self$archive$cols_y, with = FALSE]
+      if (self$surrogate$output_trafo_must_be_considered) {
+        ys = self$surrogate$output_trafo$transform(ys)
+      }
       for (column in self$archive$cols_y) {
         set(ys, j = column, value = ys[[column]] * self$surrogate_max_to_min[[column]])  # assume minimization
       }

R/AcqFunctionEI.R

@@ -81,7 +81,11 @@ AcqFunctionEI = R6Class("AcqFunctionEI",
     #' @description
     #' Update the acquisition function and set `y_best`.
     update = function() {
-      self$y_best = min(self$surrogate_max_to_min * self$archive$data[[self$surrogate$cols_y]])
+      y = self$archive$data[, self$surrogate$cols_y, with = FALSE]
+      if (self$surrogate$output_trafo_must_be_considered) {
+        y = self$surrogate$output_trafo$transform(y)
+      }
+      self$y_best = min(self$surrogate_max_to_min * y)
     }
   ),
 

R/AcqFunctionEILog.R

@@ -0,0 +1,132 @@
+#' @title Acquisition Function Expected Improvement on Log Scale
+#'
+#' @include AcqFunction.R
+#' @name mlr_acqfunctions_ei_log
+#'
+#' @templateVar id ei_log
+#' @template section_dictionary_acqfunctions
+#'
+#' @description
+#' Expected Improvement assuming that the target variable has been modeled on log scale.
+#' In general only sensible if the [SurrogateLearner] uses an [OutputTrafoLog] without inverting the posterior predictive distribution (`invert_posterior = FALSE`).
+#' See also the example below.
+#'
+#' @section Parameters:
+#' * `"epsilon"` (`numeric(1)`)\cr
+#'   \eqn{\epsilon} value used to determine the amount of exploration.
+#'   Higher values result in the importance of improvements predicted by the posterior mean
+#'   decreasing relative to the importance of potential improvements in regions of high predictive uncertainty.
+#'   Defaults to `0` (standard Expected Improvement).
+#'
+#' @family Acquisition Function
+#' @export
+#' @examples
+#' if (requireNamespace("mlr3learners") &
+#'     requireNamespace("DiceKriging") &
+#'     requireNamespace("rgenoud")) {
+#'   library(bbotk)
+#'   library(paradox)
+#'   library(mlr3learners)
+#'   library(data.table)
+#'
+#'   fun = function(xs) {
+#'     list(y = xs$x ^ 2)
+#'   }
+#'   domain = ps(x = p_dbl(lower = -10, upper = 10))
+#'   codomain = ps(y = p_dbl(tags = "minimize"))
+#'   objective = ObjectiveRFun$new(fun = fun, domain = domain, codomain = codomain)
+#'
+#'   instance = OptimInstanceBatchSingleCrit$new(
+#'     objective = objective,
+#'     terminator = trm("evals", n_evals = 5))
+#'
+#'   instance$eval_batch(data.table(x = c(-6, -5, 3, 9)))
+#'
+#'   learner = default_gp()
+#'
+#'   output_trafo = ot("log", invert_posterior = FALSE)
+#'
+#'   surrogate = srlrn(learner, output_trafo = output_trafo, archive = instance$archive)
+#'
+#'   acq_function = acqf("ei_log", surrogate = surrogate)
+#'
+#'   acq_function$surrogate$update()
+#'   acq_function$update()
+#'   acq_function$eval_dt(data.table(x = c(-1, 0, 1)))
+#' }
+AcqFunctionEILog = R6Class("AcqFunctionEILog",
+  inherit = AcqFunction,
+
+  public = list(
+
+    #' @field y_best (`numeric(1)`)\cr
+    #'   Best objective function value observed so far.
+    #'   In the case of maximization, this already includes the necessary change of sign.
+    y_best = NULL,
+
+    #' @description
+    #' Creates a new instance of this [R6][R6::R6Class] class.
+    #'
+    #' @param surrogate (`NULL` | [SurrogateLearner]).
+    #' @param epsilon (`numeric(1)`).
+    initialize = function(surrogate = NULL, epsilon = 0) {
+      assert_r6(surrogate, "SurrogateLearner", null.ok = TRUE)
+      assert_number(epsilon, lower = 0, finite = TRUE)
+
+      constants = ps(epsilon = p_dbl(lower = 0, default = 0))
+      constants$values$epsilon = epsilon
+
+      super$initialize("acq_ei_log", constants = constants, surrogate = surrogate, requires_predict_type_se = TRUE, direction = "maximize", label = "Expected Improvement on Log Scale", man = "mlr3mbo::mlr_acqfunctions_ei_log")
+    },
+
+    #' @description
+    #' Update the acquisition function and set `y_best`.
+    update = function() {
+      assert_r6(self$surrogate$output_trafo, "OutputTrafoLog")
+      assert_false(self$surrogate$output_trafo$invert_posterior)
+      y = self$archive$data[, self$surrogate$cols_y, with = FALSE]
+      if (self$surrogate$output_trafo_must_be_considered) {
+        y = self$surrogate$output_trafo$transform(y)
+      }
+      self$y_best = min(self$surrogate_max_to_min * y)
+    }
+  ),
+
+  private = list(
+    .fun = function(xdt, ...) {
+      if (is.null(self$y_best)) {
+        stop("$y_best is not set. Missed to call $update()?")
+      }
+      assert_r6(self$surrogate$output_trafo, "OutputTrafoLog")
+      assert_false(self$surrogate$output_trafo$invert_posterior)
+      constants = list(...)
+      epsilon = constants$epsilon
+      p = self$surrogate$predict(xdt)
+      mu = p$mean
+      se = p$se
+
+      # FIXME: try to unify w.r.t minimization / maximization and the respective transformation
+      if (self$surrogate_max_to_min == 1L) {
+        # y is to be minimized and the OutputTrafoLog performed the transformation accordingly
+        assert_true(self$surrogate$output_trafo$max_to_min == 1L)
+        y_best = self$y_best
+        d = (y_best - mu) - epsilon
+        d_norm = d / se
+        multiplicative_factor = (self$surrogate$output_trafo$state[[self$surrogate$output_trafo$cols_y]]$max - self$surrogate$output_trafo$state[[self$surrogate$output_trafo$cols_y]]$min)
+        ei_log = multiplicative_factor * ((exp(y_best) * pnorm(d_norm)) - (exp((0.5 * se^2) + mu)) * pnorm(d_norm - se))
+      } else {
+        # y is to be maximized and the OutputTrafoLog performed the transformation accordingly
+        y_best = - self$y_best
+        d = (mu - y_best) - epsilon
+        d_norm = d / se
+        multiplicative_factor = (self$surrogate$output_trafo$state[[self$surrogate$output_trafo$cols_y]]$max - self$surrogate$output_trafo$state[[self$surrogate$output_trafo$cols_y]]$min)
+        ei_log = multiplicative_factor * ((exp(-y_best) * pnorm(d_norm)) - (exp((0.5 * se^2) - mu) * pnorm(d_norm - se)))
+      }
+      ei_log = ifelse(se < 1e-20 | is.na(ei_log), 0, ei_log)
+      data.table(acq_ei_log = ei_log)
+    }
+  )
+)
+
+mlr_acqfunctions$add("ei_log", AcqFunctionEILog)
+

R/AcqFunctionEIPS.R

@@ -78,7 +78,11 @@ AcqFunctionEIPS = R6Class("AcqFunctionEIPS",
     #' @description
     #' Update the acquisition function and set `y_best`.
     update = function() {
-      self$y_best = min(self$surrogate_max_to_min[[self$col_y]] * self$archive$data[[self$col_y]])
+      ys = self$archive$data[, self$surrogate$cols_y, with = FALSE]
+      if (self$surrogate$output_trafo_must_be_considered) {
+        ys = self$surrogate$output_trafo$transform(ys)
+      }
+      self$y_best = min(self$surrogate_max_to_min[[self$col_y]] * ys[[self$col_y]])
     }
   ),
 

R/AcqFunctionPI.R

@@ -68,7 +68,11 @@ AcqFunctionPI = R6Class("AcqFunctionPI",
     #' @description
     #' Update the acquisition function and set `y_best`.
     update = function() {
-      self$y_best = min(self$surrogate_max_to_min * self$archive$data[[self$surrogate$cols_y]])
+      y = self$archive$data[, self$surrogate$cols_y, with = FALSE]
+      if (self$surrogate$output_trafo_must_be_considered) {
+        y = self$surrogate$output_trafo$transform(y)
+      }
+      self$y_best = min(self$surrogate_max_to_min * y)
     }
   ),
 

R/AcqFunctionSmsEgo.R

@@ -116,6 +116,9 @@ AcqFunctionSmsEgo = R6Class("AcqFunctionSmsEgo",
 
       n_obj = length(self$archive$cols_y)
       ys = self$archive$data[, self$archive$cols_y, with = FALSE]
+      if (self$surrogate$output_trafo_must_be_considered) {
+        ys = self$surrogate$output_trafo$transform(ys)
+      }
       for (column in self$archive$cols_y) {
         set(ys, j = column, value = ys[[column]] * self$surrogate_max_to_min[[column]])  # assume minimization
       }