Page 168 - AWSAR 2.0

144 || AWSAR Awarded Popular Science Stories - 2019
Let’s build intelligent machines with Hide-and-Seek
Mr. Tirtharaj Dash*
Email: tirtharaj@goa.bits-pilani.ac.in
IPrelude – Introducing the Characters
t’s a sunny afternoon in a remote village, comprising small and large houses in closed neighbourhoods, in India. For children, this is
the perfect time to play hide-and-seek (Lukaa Chhipee in Hindi).
“Ok. I am closing my eyes, you guys hide andIwillsearchforyou.Iamcountingupto3; 1... 2... 3...” calls out the boy or the “seeker”. Now, the seeker has to find one or more of his friends to win this game. He goes to one location, which is near to him, then to another. And continues till he finds one of his friends the “hider”. He continues till he locates all of them. He wins the game of hide-and-seek.
Little did I know that such a simple- yet-wonderful childhood game of hide-and- seek could be useful in building computing
machines that can learn from human experience. In scientific jargon, we call these machines “Artificial Intelligence (AI)” tools. One such tool is popularly known as “Neural Network”, which is inspired by the way a human learns from experiences. An experience is called a data instance and is defined by a set of external inputs, called “features” in AI terminology. These biologically inspired neural networks process external features and produce an action, called “output”. This output could be a category to which the given experience belongs or a quantity. The quality of the output of the neural network depends on the attributes of external features it received. An analogous example to understand this concept is to consider, your experience at a restaurant, whether the experience was good
* Mr. Tirtharaj Dash, PhD Scholar from K. K. Birla Goa Campus (BITS Pilani), Goa, is pursuing his research on “Search as a Game in Machine Learning”. His popular science story entitled “Let’s Build Intelligent Machines with Hide-and- Seek” has been selected for AWSAR Award.