diff --git a/src/run.jl b/src/run.jl
index 19ad9a5..e380d53 100644
--- a/src/run.jl
+++ b/src/run.jl
@@ -66,6 +66,9 @@ function active_learn(experiment::Dict{Symbol, Any}, data::Array{T, 2}, labels::
     set_data!(model, train_data)
     set_pools!(model, labelmap(train_pools))
 
+    classify_precision = get(experiment[:param], :classify_precision, SVDD.OPT_PRECISION)
+    debug(LOGGER, "Classify precision: $classify_precision")
+
     debug(LOGGER, "Start active learning cycle with $(experiment[:param][:num_al_iterations]) queries.")
     for i in 0:experiment[:param][:num_al_iterations]
         info(LOGGER, "Iteration $(i)")
@@ -89,7 +92,7 @@ function active_learn(experiment::Dict{Symbol, Any}, data::Array{T, 2}, labels::
         test_data, _, test_indices = get_test(split_strategy, data, pools)
         debug(LOGGER, "[TEST] Testing by predicting $(format_observations(test_data)) observations.")
         predictions = SVDD.predict(model, test_data)
-        push_evaluation!(res.al_history, i, predictions, labels[test_indices])
+        push_evaluation!(res.al_history, i, predictions, labels[test_indices], classify_precision)
         debug(LOGGER, "[TEST] Testing done.")
 
         if i < experiment[:param][:num_al_iterations]
@@ -220,14 +223,14 @@ function push_evaluation_cm!(al_history, i, cm)
     end
 end
 
-function push_evaluation!(al_history::MVHistory, i, predictions::Vector{Vector{Float64}}, labels)
-    cm = ConfusionMatrix(SVDD.classify(predictions, Val(:Global)), labels)
+function push_evaluation!(al_history::MVHistory, i, predictions::Vector{Vector{Float64}}, labels, classify_precision)
+    cm = ConfusionMatrix(SVDD.classify(predictions, Val(:Global), opt_precision=classify_precision), labels)
     push_evaluation_cm!(al_history, i, cm)
     return nothing
 end
 
-function push_evaluation!(al_history::MVHistory, i, predictions, labels)
-    cm = ConfusionMatrix(SVDD.classify.(predictions), labels)
+function push_evaluation!(al_history::MVHistory, i, predictions, labels, classify_precision)
+    cm = ConfusionMatrix(SVDD.classify.(predictions, opt_precision=classify_precision), labels)
     push_evaluation_cm!(al_history, i, cm)
     push!(al_history, :auc, i, roc_auc(predictions, labels))
     for k in [0.01, 0.02, 0.05, 0.1, 0.2]