In: Computer Science
What is model learning in computer science?
What are its uses?
what is the “Black Box”?
Model learning in Computer Science:
Model learning is a concept or sub-branch or sub-field of study in Computer Science and is a concept branch or field of study in Machine Learning (ML). Model learning, in a way, is used in ML algorithms teaching computers to perform jobs and tasks without any programs written explicitly to do so, per the instructions in it.
Model learning is used in general modeling and data-mining approaches, and specifically in model creation.
Different methods, techniques, or approaches of model learning, such as hidden Markov model learning, a task-based end-to-end model learning in Stochastic Optimization are used. And traditional model learning is the existing alternative approach we use today. etc All these approaches are used in ML. There is another area in ML called action model learning which deals with software agent's knowledge creation and modification about the actions' effects and preconditions, executed within its environment.
Model learning comes into picture or it is created applying an algorithm to data about which, an attribute of the group or class is determined to produce an algorithm or a classifier learned from the data.
The uses of model learning are in robot control, specifically in optimization based on simulation, inverse dynamics control based on approximation, to learn operational space control; it can also be applied in object-oriented model learning; there are other practical applications of model learning; and is also used in behavior-aware model learning from human-generated trajectories.
“Black Box”, with respect to model learning in ML, is a purely model-free policy optimization approach. In this approach, we do not go with learning any model at all of the random variables say, 'r'.