Machine Learning on its own is a very big topic to cover, but there is a distinct feature that divides all the models under Machine learning into two groups, Supervised Machine Learning and Unsupervised Machine Learning. That distinct feature is the availability of labelled data. Supervised Learning is the process of learning from data which is distinctly marked using labels. In simple terms, labels serve as a guiding path for the learning process. Labels are like a teacher to the model. The labels have a direct correlation with the prediction that the model makes as it has been trained on those labelled features. This is similar to training a child to identify an apple, where the instructor tells that apple has a distinct shape with the color red. Here, the color, red acts as a label for the data(apple). Now the child can easily identify what is an apple, as he has to only search for the color red. On the other hand, Unsupervised Learning Models do not include labels so the algorithm or in this case, the child has to figure out how to identify an apple without knowing its color. Training, building, modifying and maintaining a Supervised Learning model is much easier.
Many algorithms fall under this category such as
- Support Vector Machines
- K-Nearest Neighbors
- Decision Trees
There are many more, but for simplicity’s sake I have listed only a few.
A developer uses a Supervised Learning model when he has access to a well made, labelled dataset. Supervised Learning is the easiest subset of Machine learning algorithms.
In the next blog we will be discussing about the prerequisites of Machine Learning. Until then, thank you and have a nice day! 🙂