Is it true that, like decision trees, neural networks can use any data type for independent attributes?

• A. True
• B. Cannot be determined
• C. False
• D. Not applicable

Answer: (C)

Neural networks work with numeric inputs. Everything fed into a neural network must be represented as numbers in a fixed-size numeric tensor, because the training process relies on arithmetic operations. When you have categorical attributes or text, you don’t feed them in raw. They must be converted into numbers through encoding schemes such as one-hot encoding for categories, ordinal encoding, or embeddings for more complex categorical or text data. Images and audio are already numeric or are converted into numeric feature representations, but even then the data must be prepared in a suitable numeric form.

Decision trees can split directly on discrete values and handle a mix of data types more naturally without extensive preprocessing, which is why the statement that neural networks can use any data type directly is not true. So the best answer is that this is false.