Original article was published by Srikanth Mandru on Artificial Intelligence on Medium
How does machine learning work?
Basically, in the programming, we provide the set of rules to make the machine perform a particular task for us. But, what if there are many rules that we cannot incorporate in our code for some task. For Example, consider a simple example of “Human Identification” from an Image. In order to identify the human, in the general programming approach, we need to code all the rules related to face features (eye size, ears size, hairstyles), body size, dimensions of the body, etc. It is a lot of work to include those features to identify a human. What if we could make the program to learn those features and store them for future predictions. This is the key idea behind Machine learning. As a fun fact, the machine is not learning anything by itself, it’s the program that makes the machine intelligent.
So, How does the machine learning algorithm really learn? The basic ingredients for learning are the “Data” and “algorithm”. Given the data, the computer’s algorithm learns the features that are important for the task it has been trained. Once we have those learned features, we can use that machine learning algorithm to make predictions on future data.