Study Guide - Machine Learning

Which of these best describes the difference between traditional programming and machine learning?

• Traditional programming uses inputs with algorithms to calculate outputs, whereas machine learning uses inputs and outputs to find an algorithm.

• Traditional programming is easier than machine learning.

• Traditional programming can use iteration or recursion, but machine learning can only use iteration.

• Traditional programming can be object-oriented, while machine learning can't.

Inputs used for training a machine learning model are called ___.

Answer explanation

Inputs used for training a machine learning model are called features.

Outputs used for training a machine learning model are called ___.

Answer explanation

Inputs used for training a machine learning model are called labels.

Pairs of inputs and outputs used for training a machine learning model are called ___.

Answer explanation

Pairs of inputs and outputs used for training a machine learning model are called examples.

Which of these libraries are used for machine learning?

• TensorFlow

• NumPy

• Pyplot

• Pip

Answer explanation

In Python, there are several machine learning libraries, such as TensorFlow, PyTorch, and Theano.

Which of the following are required for a machine learning model to compile properly? Select all that apply.

• Training examples

• A loss function

• An optimizer function

• Test examples

Answer explanation

To compile, a machine learning model needs three things: training examples, a loss function, and an optimizer function.

Which of the following are Keras layers? Select all that apply.

• Dense

• Flatten

• Sequential

• Adam

Answer explanation

Keras includes several machine learning layers, including Dense and Flatten.

Which of the following best improves a machine learning model?

• Using more training examples

• Training the model over more epochs

• Adding more layers

• Adding more neurons per layer

Answer explanation

Although each of these actions can improve a model under the right circumstances, using more training examples will almost always improve optimization and reduce loss.

Which of these is not one of the three layers used in machine learning models?

• input

• hidden

• output

• activation

Answer explanation

The three layers used in machine learning models are input, hidden, and output.

Which of these best describes an activation function?

• A function which helps the model decide which outputs to keep, and which to drop

• A function which activates the neural nodes of a layer

• A function which measures how accurate a model is

• A function which improves the performance of a model