Generative modeling is the use of artificial intelligence (AI), statistics and probability in applications to produce a representation or abstraction of observed phenomena or target variables that can be calculated from observations.
Generative modeling is used in unsupervised machine learning as a means to describe phenomena in data, enabling computers to understand the real world. This AI understanding can be used to predict all manner of probabilities on a subject from modeled data.
In unsupervised machine learning, generative modeling algorithms process volumes of training data and make reductions about the data into its digital essence. These models generally are run on neural networks and can come to naturally recognize the natural distinctive features of the data. The neural networks take these reduced fundamental understandings of real world data and then use them to model data that is similar or indistinguishable from real world data.
An example of a generative model might be one that is trained on collections of images from the real world in order to generate similar images. The model might take observations from a 200GB set of images and reduce them into 100MB of weights. Weights can be thought of as reinforced neural connections. Through increased training, an algorithm comes to produce more realistic images.
Generative modeling contrasts with discriminative modeling, which identifies existing data and can be used to classify data. Generative modeling produces something; discriminative modeling recognizes tags and sorts data. In the above example, a generative model can be enhanced by a descriptive model and vice-versa: this is done by having the generative model trying to fool the discriminative model into believing the generated images are real. Through successions of training, both become more sophisticated at their tasks.