ML metaphor question
So in this video, the concept of ML training is likened to a robot shooting an arrow at a target over and over until the robot learns how to hit the target perfectly. In the real world, this target often changes after X time interval. Does existing ML techniques address this dynamic nature of real-world data?
Great question! The courses that you will find later on in the program teach you that ML models cannot be productionized and left alone. You constantly need to test whether the model is still valid and the new data coming in has the same characteristics as the data you initially trained your model on. This is an important task handled by data scientists and ML Ops engineers.