TypeError: can't pickle _thread.RLock objects when adding a Neural Network to an Stacking Ensemble

I am currently trying to build a stacking ensemble that consists of both "standard models" and a neural network. The ensemble contains Random Forest, XGBoost, SVM and Catboost. But as soon as I add the neural network I get the error "TypeError: can't pickle _thread.RLock objects". I have tried different versions of Tensorflow (2.0.0, 2.3.0, 1.14, 1.13) but that did not solve the problem. I hope someone can help me with this case.

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

from sklearn.model_selection import train_test_split
from sklearn.model_selection import StratifiedKFold

from sklearn.preprocessing import StandardScaler
from sklearn.preprocessing import MinMaxScaler
from sklearn.preprocessing import RobustScaler

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation, Flatten
from tensorflow.keras.optimizers import *

rs = 23

dataset = pd.read_csv(url,sep='|')

x = dataset.drop('fraud', axis=1)
y = dataset.fraud

x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.3, stratify=y, random_state=rs)

scaler = StandardScaler()
scaler.fit(x_train)
x_train = scaler.transform(x_train)
x_test = scaler.transform(x_test)

Classifiers

from sklearn.ensemble import RandomForestClassifier
from sklearn.svm import SVC
from xgboost import XGBClassifier
from catboost import CatBoostClassifier

cb_clf = CatBoostClassifier(border_count=14, depth=4, iterations=600, l2_leaf_reg=1, silent= True, learning_rate= 0.02, thread_count=4, random_state=rs)
rf_clf = RandomForestClassifier(n_estimators = 700, criterion = "entropy", min_samples_leaf = 1, min_samples_split = 2, random_state = rs)
svc_clf = SVC(kernel = 'linear', C = 40, random_state = rs)
xg_clf = XGBClassifier(booster="gblinear", eta=0.5, random_state=rs)

DNN

x_train_dnn = np.array(x_train)
x_test_dnn = np.array(x_test)
y_train_dnn = np.array(y_train)
y_test_dnn = np.array(y_test)

def build_nn():

    dnn = Sequential()

    dnn.add(Dense(128, activation='relu', kernel_initializer='random_normal', input_dim=10))
    dnn.add(Dense(128, activation='relu', kernel_initializer='random_normal'))
    dnn.add(Dense(1, activation='sigmoid', kernel_initializer='random_normal'))
    dnn.compile(optimizer ='adam',loss='binary_crossentropy', metrics =['accuracy'])
  
    return dnn

dnn_clf = keras.wrappers.scikit_learn.KerasClassifier(
                            build_nn,
                            epochs=500,
                            batch_size=32,
                            verbose=False)

dnn_clf._estimator_type = "classifier"

from sklearn.ensemble import StackingClassifier
from sklearn.linear_model import LogisticRegression

estimators = [("Random Forest", rf_clf),
              ("XG", xg_clf),
              ("SVC", svc_clf),
              ("Catboost", cb_clf),
              ("DNN", dnn_clf)]

ensemble = StackingClassifier(estimators=estimators, n_jobs=-1, final_estimator=LogisticRegression())

Fitting the Ensemble causes the error

ensemble.fit(x_train, y_train)#fit model to training data
ensemble.score(x_test, y_test)#test our model on the test data

The above exception was the direct cause of the following exception:

TypeError                                 Traceback (most recent call last)
<ipython-input-14-1c003d476ea2> in <module>()
----> 1 ensemble.fit(x_train, y_train)#fit model to training data
      2 ensemble.score(x_test, y_test)#test our model on the test data

6 frames
/usr/lib/python3.6/concurrent/futures/_base.py in __get_result(self)
    382     def __get_result(self):
    383         if self._exception:
--> 384             raise self._exception
    385         else:
    386             return self._result

TypeError: can't pickle _thread.RLock objects