What is target output?
A target output is the true output or labels on a given dataset. The function that maps the input to its correct labels is called the target function. Therefore, the underlying goal of many machine learning methods is to produce a function that matches the target function as close as possible without giving up generalizability. The target output can be used to compare the predictions of a model and determine its accuracy.
Example of Target Output
Consider a neural network that classifies images. This could be a classification like dog vs cat vs bird, +1 vs -1, and so on. Lets say our network classifies dogs, cats, and birds and that we have (in order) the following input:
We need to classify each of these images to know if our neural network is correctly classifying them as well. In order, we classify each of these as follows:
[cat, cat, dog, bird, dog]
This array is called the “target array” and contains the target outputs. If our network were to predict [cat, dog, cat, bird, dog], this would be called the “predicted array” which contains the predicted outputs.
When the predicted and target arrays are compared, we see that the model correctty classified ⅗ of the inputs. This model, thus, has an accuracy of 60% on this data set.