Binary classification
Introduction
Binary classification is a type of supervised machine learning task in which an algorithm is trained to classify an input into one of two possible categories, often represented as "positive" and "negative". To train its algorithm, it uses an existing dataset which contains inputs and their labels indicating which category each belongs in. Once trained, this algorithm can be applied to new inputs to predict their labels accurately.
A binary classification problem seeks to discover a function that accurately maps inputs to one of two possible categories. This boundary typically depicts this boundary as a decision boundary, dividing input space into two regions - one for each category - where points on one side are classified as belonging in one category while those on the other side fall under another classification.
Binary classification often relies on logistic regression. This algorithm employs a logistic function to model the likelihood that an input belongs to a certain category, with its output thresholded for final prediction accuracy.
Support Vector Machines (SVMs) are another popular algorithm for binary classification. SVMs try to identify the optimal decision boundary that maximizes margin, or distance, between its closest points and those of each category. This produces a boundary which is more resilient to noise and outliers.
Random Forest is another ensemble learning algorithm that utilizes multiple decision trees to make predictions. It works by training multiple trees on different subsets of the training data, then averaging their predictions together for a final prediction.
Explain Like I'm 5 (ELI5)
Binary classification" is like when a teacher wants to know if you prefer apples or oranges more. They give both fruits and ask you to pick one, then mark it down on paper. The teacher does this with many other students so they can determine whether new students will prefer apples or oranges better.
An alternative way of looking at it is like sorting objects into two separate boxes, one labeled "yes" and the other labeled "no". The computer does this by taking pictures of the items and deciding which box it should put them in based on what it sees.