phew

Project.toml

@@ -19,7 +19,6 @@ MLJEnsembles = "50ed68f4-41fd-4504-931a-ed422449fee0"
 MLJFlux = "094fc8d1-fd35-5302-93ea-dabda2abf845"
 MLJModelInterface = "e80e1ace-859a-464e-9ed9-23947d8ae3ea"
 MLUtils = "f1d291b0-491e-4a28-83b9-f70985020b54"
-MultivariateStats = "6f286f6a-111f-5878-ab1e-185364afe411"
 Parameters = "d96e819e-fc66-5662-9728-84c9c7592b0a"
 PkgTemplates = "14b8a8f1-9102-5b29-a752-f990bacb7fe1"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"

experiments/benchmarking/benchmarking.jl

@@ -8,7 +8,7 @@ function default_generators(;
     nsamples::Union{Nothing,Int}=nothing,
     nmin::Union{Nothing,Int}=nothing,
     reg_strength::Real=0.5,
-    dim_reduction::Union{Nothing,Int}=nothing,
+    dim_reduction::Bool=false,
 )
 
     @info "Begin benchmarking counterfactual explanations."

experiments/mnist.jl

@@ -13,7 +13,7 @@ counterfactual_data.generative_model = vae
 
 # Dimensionality reduction:
 maxout_dim = vae.params.latent_dim
-counterfactual_data.dt = MultivariateStats.fit(PCA, counterfactual_data.X; maxoutdim=maxout_dim);
+counterfactual_data.dt = MultivariateStats.fit(MultivariateStats.PCA, counterfactual_data.X; maxoutdim=maxout_dim);
 
 # Test data:
 test_data = load_mnist_test()

experiments/setup_env.jl

@@ -23,12 +23,12 @@ using MLJ: TunedModel, Grid, CV, fitted_params, report
 using MLJBase: multiclass_f1score, accuracy, multiclass_precision, table, machine, fit!, Supervised
 using MLJEnsembles
 using MLJFlux
-using MultivariateStats
 using Random
 using Serialization
 using Statistics
 
 import MPI
+import MultivariateStats
 
 Random.seed!(2023)
 

src/penalties.jl

@@ -4,7 +4,6 @@ using Distances
 using Flux
 using Images: assess_ssim
 using LinearAlgebra: norm
-using MultivariateStats
 using Statistics: mean
 
 """