DYNAMISM(E) - Biannual Student Magazine June-2017 | Page 15

need of yet another learning algorithm when we already have linear regression and logistic regression. The reason is that a simple logistic regression together with adding in quadratic or the cubic features to provide a reliable answer, will end up with millions of features and that’s just too much. This is not a good way to do it. We need something better and that’s were Neural Network steps in. Another reason for the emergence of Neural Networks is the emergence of big and fast computer that makes large scale Neural Networks to work economically. Neural Networks are actually very effective state of the art technique for modern day machine learning applications and get them to work well on problems. Without diving into the technical details and avoiding mathematical definition, let us understand how the neural network looks. At a very simple level, neurons are basically computational units that take a number of inputs (dendrites) as electrical inputs (called 15 “spikes”) and does some computation and then are channeled to outputs (axons) to other nodes or to other neurons in the brain. An example of neural network is given in figure 1. Armed with above knowledge, one is in a position to understand the terms machine learning and Neural Network. It also emerges from above that the scope of machine learning seems to be unlimited. In general, there is wide scope for people with good knowledge of mathematics and particularly IT and electrical engineers to cash in on this opportunity. Management knowledge would be an added advantage. DYNAMISM(E)