Why Probability is the Foundation of Machine Learning!!

Act I: Understanding the Real World

In our day-to-day life, people do many miraculous things such as cooking Italian pasta, catching balls that are thrown at them, driving cars, fixing a machine, calculating accounts to spend on a new pair of shoes. People do this every day without sweating. All of this requires a predetermined amount of concentration and calculation. While cooking Italian Pasta, one must consider the amount of salt or the quality of the cheese. If one ingredient differs from the previous cooking, it will taste vastly different. When someone needs to catch a ball that is thrown at them, he/she sees and understands the trajectory the ball is coming in. The velocity, the time, the position, everything is calculated precisely to catch that ball. People do all these without even thinking. But all of these occurrences require people to calculate what has happened and what is going to happen. That is where probability comes in.

Probability refers to the ratio of the number of occurrences for a situation and the total number of occurrences. All of the real-world things that people do require some amount of probability to be calculated to predict the possible future. Probability is just a fancy mathematical term to identify if an occurrence is more likely to happen. If the value of the probability of occurrence is higher than the value of the probability of another occurrence then the previous occurrence is more likely to happen.

Act II: Understanding Machine learning

From the early 50’s till this day, Computers have been helping us in various tasks in various ways. The first computer that was built by Turing was built to predict outcomes and break the encryption system. And it was successful in doing so. The use case of computers for us has not changed that much. From that time, we have been using computers to ease up our work from calculating accounting info, writing documents to sharing ideas with complete strangers over the internet. But nowadays we have again gone to the same task, predicting the future.

When the task is to predict the future, computers try to understand the situation that has happened. But computers are just machines that can compute. They can not understand the real world as is. So, we have to numberize (converting to numeric values) everything so that computers can understand the situation. It does so by calculating the ratio of the occurrences.

Act III (Final act): Use of probability in Machine learning

When the computers try to understand the situation it only compares the values of the ratio of the occurrences. It is the Probability. So when the computer is trying to understand the situation and trying to predict the possible future, it is only calculating probabilities and comparing them for each situations.