The model prediction and the explanation of deep learning framework (VAE-MLP) , including counterfactual explanations and layerwise relevance propagation