Month: July 2019

My First Post In Machine Learning

It was 2007 when I first heard of something like this called ‘artificial intelligence’. for me it was then a robotic arm controlled by a program written in BASIC language. Later i realised it has something related to the machines who started to think on themselves and trying to control the humans, Yeah, just like The terminator. Thankfully machines have not rebelled against human like it was shown in the movie but yes somehow machines can somewhat think and learn! How !!

The type of learning where you learn from your experience is called Inductive learning, e.g. when you take a test you become better at taking test when you take the test again and again. Doctors become better at diagnosing and treating patients with experience

Well the thinking part is not much developed but the learning part has. Consider this, we go to doctor, he checks the eyes, tongue and pulse rate, then suggest two medicines. now Imagine doing visiting a computer who claims to be a doctor. The camera in front of computer clicks the picture of tongue and eyes and then the straps take the reading of pulse. (well thats not big deal). After few minutes it prescribes the same two medicines (or same composition). Thats a perfect example of machine learning.

So a washing machine learns to wash your clothes automatically is that machine leaning ! A machine learning is application of artificial intelligence where a system / a program / a machine which runs on a program improves its results through its experience of previously delivered results. A computer program takes in input I processes it provide the output O but wait! it takes the output learns from it and repeats the process to improve the output O. Thats machine learning. So what does it actually learn ? It learns whatever it is programmed to do so.

Lets consider some practical examples :

  1. Gmail : Gmail automatically puts some of the emails in the spam folder, whereas it let other emails land in the Inbox.
  2. Loan : Some people don’t get loan/ credit card because they had poor history, because banks have anticipated who all are the potential defaulters.
  3. Amazon : We buy something and automatically it suggests another product that could be useful to buy with the first one. How can it do with millions of products. That M.L. for you.
  4. Some websites ask you to verify that your nota robot by identifying a traffic light, cat dog or what not in the image, How does the website know if it a traffic light, that machine which learnt from other such images.
  5. Handwriting recognition softwares does reasonably well in identifying scribbly written words.