Do you really know the difference between Test and Validation Datasets?

Many people don't really know the difference between test and validation. In Machine Learning these two words are often used improperly, but they indicate two very different things. Even literature sometimes reverses the meaning of these terms. When training a model the dataset is usually divided into a train set, a validation set and a test set but…why are the last two sets needed? Keep reading, you will find your answers.