AI Discovery 2023 - Semester 1 Final

What type of machine learning was primarily used in the fruit classification project?

• Reinforcement Learning

• Supervised Learning

• Unsupervised Learning

• Semi-supervised Learning

What machine learning language did we use to demonstrate the the fruit classification project?

• Python

• PyTorch

• TensorFlow

• Keras

What is a key step when preparing data for training a machine learning model?

• Overfitting the model

• Increasing the data size by making up data

• Labeling the data

• Training the model with the data

What does the accuracy of the model measure in a classification model?

• The model’s ability to predict negative outcomes

• The model’s ability to predict positive outcomes

• The time it takes to train the model

• The proportion (the amount) of times the model got the right answer

In the context of the fruit classification project, what would ‘feature engineering’ involve?

• Selecting the most expensive machine learning model

• Making the model predict answers faster

• Changing the color scheme of the fruit images

• Altering (editing) or creating new input attributes (features) for the model to use in its predictions

Answer explanation

Remember this is the process used to support which characteristics will be used to identify the fruit. For example, in the fruit classification project, we used the following 3 features to train the model. We could have used different or added more if we wanted to improve the performance of the model.

What is a common technique used to assess a model’s performance on different subsets (parts) of the training data?

• Cross-validation

• Undercutting

• Data mining

• Data transmission

Answer explanation

Remember this is used to evaluate the model when it sees unseen data. It’s when we swap out the data with different variations and combinations of the data to ensure the model is not overfitted. You divide the available data into different datasets and train the model on that only to observe how it performs.

In the fruit classification project, what kind of data will be inputted into the model for the model to predict an answer? Hint: don’t think about the training or test data.

• Text data

• Image data

• Audio data

• None of the above

Answer explanation

Remember text is what we used to build the training and test data in tabular format. However, the data that would be used by the user to input the model would be images (of fruits). The images is what we will feed the model and the model will output what we program it to do. For example, the name of the fruit or category or something as simple as is this a kiwi fruit, yes or no.

In the fruit classification project, what features did we use?

• Weight, height, and color

• Name, shape, and can you eat it with a spoon

• Weight, color, and texture

• Color, seeds, and itchy

Answer explanation

Note: Defining characteristics is easy. Finding the right characteristics that can be used to train a model to predict something is the most important thing. Not all characteristics are equal.

Which of the following best describes a ‘Decision Tree Classifier’?

• A model that predicts based on the majority rule of nearest neighbors

• A model that uses nodes to make decisions (think organizational structure)

• A model that uses linear relationships to make decisions (think about a straight line)

• A model that splits the data and choose either A or B

After completing the model, what is the next step before making it available to other people to use it?

• Deleting the model

• Saving the model

• Restarting the entire project

• Change the programming library to something else