Introduction to Machine Learning Lesson 9: Semi-Supervised Learning Lesson 9 Quiz

Which of these situations is ideal for your semi-supervised learning application? A You plan to use several small labeled datasets to train the system. B You plan to use a small, labeled dataset to predict the output for a much larger unlabeled dataset. C You plan to use several large unlabeled datasets to train the system. D You plan to use one large unlabeled dataset to train the system.

You plan to use a small, labeled dataset to predict the output for a much larger unlabeled dataset.

What is a pseudo-label? A A label generated during self-training that cannot be verified as accurate. B A label that does not answer the question the machine is tasked with answering. C A label manually created or assigned by a human. D A label calculated from one or more other features.

A label generated during self-training that cannot be verified as accurate.

In what way is semi-supervised learning sometimes better than supervised learning? A It is less prone to prediction errors. B It uses labeled data. C Adequate training data is easier to get. D It trains more quickly.

Adequate training data is easier to get.

In order to use labeled data to make assumptions about unlabeled data, you need to be confident that all the data comes from the same place, so it is likely to have the same properties. Which assumption is that an example of? A Continuity B Manifold C Adjacency D Cluster

Continuity

In what way is semi-supervised learning sometimes better than unsupervised learning? A It allows for clustering data into groups. B It enables the model to label some of the unlabeled data. C Adequate training data is easier to get. D It uses unlabeled data.

It enables the model to label some of the unlabeled data.