Multiclass Classification Quiz and Flashcards for telmef

Study Notes

Multiclass Classification

Multiclass classification is a supervised machine learning technique used to predict to which of multiple possible classes an observation belongs.

Iterative Process

The iterative process involves training, validating, and evaluating the model, with a subset of the training data held back to validate the trained model.

Multiclass Classification Algorithms

Multiclass classification algorithms calculate probability values for multiple class labels, enabling a model to predict the most probable class for a given observation.

Training a Multiclass Classification Model

Two kinds of algorithms can be used to train a multiclass classification model: One-vs-Rest (OvR) and Multinomial algorithms.

One-vs-Rest (OvR) Algorithms

OvR algorithms train a binary classification function for each class, calculating the probability that the observation is an example of the target class compared to any other class.
Each function produces a sigmoid function that calculates a probability value between 0.0 and 1.0.
The model predicts the class for the function that produces the highest probability output.

Multinomial Algorithms

Multinomial algorithms create a single function that returns a multi-valued output, a vector containing the probability distribution for all possible classes.
The output is a vector with a probability score for each class, totaling up to 1.0.
An example of this kind of function is a softmax function.

Evaluating a Multiclass Classification Model

A multiclass classifier can be evaluated by calculating binary classification metrics for each individual class.
Alternatively, aggregate metrics can be calculated, taking all classes into account.
The confusion matrix for a multiclass classifier shows the number of predictions for each combination of predicted and actual class labels.
Metrics for each individual class can be determined from the confusion matrix.
Overall accuracy, recall, and precision metrics can be calculated using the total of the TP, TN, FP, and FN metrics.
The overall F1-score is calculated using the overall recall and precision metrics.

Description

This quiz covers multiclass classification, a supervised machine learning technique used to predict the class of an observation. It involves training, validating, and evaluating a model to calculate probability values for multiple classes.