RoboGrok Robotics 2 IK Quiz

Robotics 2
Machine Learning Supervised
Quiz

A Support Vector Machine will usually work better for data that has more dimensions, or 'features'.

Your ID Code:

Question 3: Which of these describes how an Artificial Neural Network works?

Question 2: Which of these describes how the Support Vector Machine algorithm works?

Question 1: What do we call the number that the supervisor places on a datapoint to indicate which class it belongs to?

Label

Cluster

Value

Indicator

Support Vector Machine finds the line with most curvature between the data points in two classes.

Support Vector Machine finds groupings in data by starting with a 'seed', checking the distance between each data point and the seed, then assigning each point to the nearest seed.

Support Vector Machine consists of many simple elements, each of which receives an input, multiplies it by a 'weight', adds up all the inputs, and limits the result value.

Support Vector Machine finds a hyperplane that optimally separates the two classes of data by equalizing the distance between the hyperplane and the closest point in each class.

Question 4: What does it mean for data to be 'separable'?

Data is separable if it is possible to find a straight line, plane, or hyperplane between the data classes.

Question 5: True or false: data must be separable for Support Vector Machine to work

Data is separable if it consists of fewer than 3 classes.

True

False

Data is separable if none of the data points have been mislabeled by the supervisor.

Data is separable if it is possible for a machine learning algorithm to learn to separate the classes.

Question 7: What is one reason we might we choose to use fewer neurons in an Artificial Neural Network?

Question 8: What is one reason we might choose to use Support Vector Machine rather than an Artificial Neural Network?

Question 9: What is one reason we might we choose to use an Artificial Neural Network rather than Support Vector Machine?

Question 10: True of False: Support Vector Machine won't work if we have more than 2 classes of data.

Artificial Neural Network finds the line with most curvature between the data points in two classes.

Artificial Neural Network finds groupings in data by starting with a 'seed', checking the distance between each data point and the seed, then assigning each point to the nearest seed.

Artificial Neural Network consists of many simple elements, each of which receives an input, multiplies it by a 'weight', adds up all the inputs, and limits the result value.

Artificial Neural Network finds a hyperplane that optimally separates the two classes of data by equalizing the distance between the hyperplane and the closest point in each class.

False

True

Question 6: True or False: data must be separable for Artificial Neural Networks to work

Using fewer neurons allows us to classify data with more dimensions, or 'features'

Using fewer neurons can allow an Artificial Neural Network to separate more complex data

Using more neurons can cause 'overfitting', where the separating line between the classes has too many curves

Using more neurons can cause the learning process to fail if there isn't enough training data.

A Support Vector Machine works better than Artificial Neural Networks when the data are not labeled.

A Support Vector Machine can be more efficient than Artificial Neural Networks, because it only needs to 'see' each data point one time during the training process.

A Support Vector Machine can find a separating line, or plane, with lots of curves; an Artificial Neural Network always finds a straight line or plane.

An Artificial Neural Network is often better at classifying data that is not separable, because it is able to find a separating line with curves.

An Artificial Neural Network is usually much less computationally-intensive than Support Vector Machines.

An Artificial Neural Network can be used to classify more than 2 classes of data, whereas Support Vector Machines can only be used with 2 classes.

An Artificial Neural Network is more efficient than Support Vector Machines, because the Artificial Neural Network only needs to 'see' each data point one time during the training process.

False

True