In today's world, we are generating lots n lots of data. By 2025, it projected that we will have generated 175 zettabytes of data. But, even though we have so much of data, why is it so difficult to get right data for right model?. Because vast majority of the data being generated is not labeled, and most of the models need labeled data, i.e they use supervised learning. Labeling is an expensive and time consuming job.