Types of Machine Learning

What are the three types of machine learning?

• Supervised learning, Unsupervised learning, Reinforcement learning

• Supervised learning, Association learning, Regression learning

• Supervised learning, Unsupervised learning, Repetitive learning

• Supervised learning, Unsupervised learning, Predictive learning

Why is there a need for both training and test data in machine learning?

• To increase the size of the database.

• To ensure the model can generalize beyond the training data.

• To make the training process faster.

• To enhance the graphical representation of data.

Which type of machine learning are you using when training a model with labeled images of apples and tomatoes?

• Supervised learning

• Unsupervised learning

• Reinforcement learning

• Semi-supervised learning

Answer explanation

Training a model with labeled images of apples and tomatoes falls under supervised learning where the algorithm learns from labeled data to make predictions.

What is done with the remaining data after a machine learning model is trained?

• It is discarded

• It is used to train the model further

• It is used to test the model

• It is used to create another model

Answer explanation

The remaining data after training is used to test the model's performance and evaluate its accuracy.

What is bias in the context of machine learning models?

• When the model is completely accurate and fair

• When the model requires additional data to function

• When the model favours some things and deprioritises or excludes others

• When the model operates independently without any human intervention

Answer explanation

Bias in machine learning models refers to favouring some things while deprioritizing or excluding others, leading to unfairness or inaccuracies. e.g A facial recognition system that is less accurate in recognising people with certain skin tones

Why might an AI application that predicts jobs based on employment data from 1960 exhibit societal bias?

• Because the data from 1960 accurately reflects the current job market.

• Because the data from 1960 is too recent to show any significant trends.

• Because the data from 1960 may not reflect the current societal norms and values.

• Because the data from 1960 reflects biases that existed in society at that time.

Answer explanation

The correct choice is because the data from 1960 reflects biases that existed in society at that time, which can lead to societal bias in the AI application.

Can societal bias lead to data bias?

• Yes, always

• No, never

• Yes, it can

• No, it is unrelated

Answer explanation

Societal bias can influence the data collected, leading to biased results. Therefore, societal bias can indeed lead to data bias.

When designing a machine learning application, whose backgrounds, experiences, and opinions should be considered to reduce bias?

• The developers of the ML application

• The stakeholders of the project

• People with different backgrounds, experiences, and opinions

• The end-users of the ML application

Why is it important to use large and representative sets of data to train ML models?

• To increase the speed of training

• To make the model less accurate

• To reduce bias in the model

• To make the training process more expensive

Answer explanation

Using large and representative sets of data helps reduce bias in the model by capturing a wider range of patterns and relationships, leading to more accurate predictions.