12 Deep Learning Interview questions you should not be missed (Part 3)

9. When the input image size is doubled, how many times does the number of CNN parameters increase? Why?

This is a very misleading question for candidates because most people will be thinking in the direction of the question of how many times the number of CNN parameters will increase. However, let’s take a look at CNN’s architecture:

We can see that the number of parameters of the CNN model depends on the number and size of the filters, not on the input image. Therefore, doubling the size of the image does not change the number of parameters of the model.

10. What are some ways to deal with imbalance data?

This is a question that tests the candidate’s approach to problems with real data. Usually, the actual data will vary greatly in terms of properties as well as the amount of data from the sample data sets (standard data sets do not need adjustment). With the actual data set, there can be a case where data imbalance, means that data is unbalanced between classes. We can now consider the following techniques:

• Choose the right metric to evaluate the model: with an unbalanced dataset, using accuracy to evaluate is a very…