Please explain, in light of the aforementioned three reasons, why it is preferable to have accessors for private types as opposed to making them public.

Explain the use of wrapper classes for non-object data types, so those wrapper methods can be implemented alleviating the issues of mixed non-object values with object-related values.

Explain how to utilize wrapper classes for non-object data types to avoid mixed non-object and object values.

Question 5 Suppose you want to use a GNN to classify the nodes of a social network (for example, the classes may be: "Democrat", "Republican", “Undecided", "Other"). Suppose that you know the correct label (i.e., the political affiliation) for some of the nodes. Additionally, you know some additional attributes (gender, age, education level) for all nodes. Which of the following statements is/are correct? The known node attributes, together with the political affiliation data, can be used to form the embedding of each node at layer-O. O The training data should consist of all nodes for which we know the political affiliation. O The larger the number of layers (depth) of the network, the more susceptive the network is to overfitting. O The network can be trained with the cross-entropy loss function we presented in class.