An AI machine learning method that trains a neural network by example. Supervised learning feeds the network predefined inputs in both the training and fine tuning stages of the model. Supervised learning was common in early machine learning. However, todays large models use one of the three following See
AI training.
Unsupervised = No Labels
There are no labels on the input
in unsupervised learning. The AI neural network detects patterns by recognizing common attributes. Unsupervised learning is less widely used than supervised learning, but both methods are used to train similar models.
Self-Supervised = Generated Labels
Used heavily in today's large models, self-supervised learning creates its own labels.
Semi-Supervised = Labeled and Unlabeled
Employs labeled and unlabeled datasets. Labeling a small amount of input saves a lot of human time, and it can provide enough samples to let the model identify the larger unlabeled set. See
AI model,
machine learning and
deep learning.