Using Pretrained Models

Sometimes the optimization problem we have was already solved by others but on different data.

So we can use a pretrained model to use their learnings on their data and just improve it with our specific case.

It is like learning to walk vs. learning to run. The difference is not that significant but the outcome is very different.

So a training is necessary for this case. But we can build on top of how the model learned walking to teach it how to run.