Semi-supervised learning basics in ML Python - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is semi-supervised learning?

Semi-supervised learning is a type of machine learning that uses a small amount of labeled data along with a large amount of unlabeled data to train models. It helps improve learning when labeling data is expensive or time-consuming.

beginner

Why use semi-supervised learning instead of supervised learning?

Because labeling data can be costly or slow, semi-supervised learning uses many unlabeled examples to help the model learn better patterns without needing as many labeled examples.

intermediate

Name two common methods used in semi-supervised learning.

Two common methods are: 1) Self-training, where the model labels unlabeled data and retrains itself, and 2) Graph-based methods, which use connections between data points to spread label information.

beginner

What role does unlabeled data play in semi-supervised learning?

Unlabeled data helps the model understand the overall structure and distribution of the data, which improves its ability to classify or predict even with few labeled examples.

beginner

Give a real-life example where semi-supervised learning is useful.

In medical imaging, labeling images requires expert doctors and is expensive. Semi-supervised learning can use a few labeled images and many unlabeled ones to build good models for diagnosis.

What type of data does semi-supervised learning use?

A Both labeled and unlabeled data

B Only labeled data

C Only unlabeled data

D No data at all

Semi-supervised learning combines a small amount of labeled data with a large amount of unlabeled data.

Why is semi-supervised learning helpful?

A It ignores unlabeled data

B It only uses labeled data

C It requires more labeled data than supervised learning

D It reduces the need for many labeled examples

Semi-supervised learning helps when labeled data is scarce by using unlabeled data to improve learning.

Which method is NOT typically used in semi-supervised learning?

A Graph-based methods

B Reinforcement learning

C Self-training

D Label propagation

Reinforcement learning is a different type of learning and not a common semi-supervised method.

In semi-supervised learning, unlabeled data helps the model by:

A Providing information about data structure

B Replacing labeled data completely

C Confusing the model

D Being ignored during training

Unlabeled data helps the model learn the overall shape and distribution of data.

Which scenario is a good fit for semi-supervised learning?

A When only unlabeled data is available

B When there is plenty of labeled data

C When labeled data is expensive and unlabeled data is abundant

D When no data is available

Semi-supervised learning is best when labeled data is limited but unlabeled data is easy to get.

Explain what semi-supervised learning is and why it is useful.

Think about how using both labeled and unlabeled data helps when labels are hard to get.

Describe two common methods used in semi-supervised learning and how they work.

One method lets the model label data itself; the other uses connections between data points.