Imagine 10,000 years in the future. Two archaeologists are digging a site, and they find the remains of human bones. They take those remains to the lab, perform tests on them, and try to reconstruct the habits and lifestyle of that human. Say those bones belong to me. They are trying to find out my story, how I lived, what I ate, where I traveled, etc. Once they find all the info they need, they toss it into a museum where they store all other such ancient human remains they find. I was essentially one data point to them in trying to understand how humanity evolved over time. Or I can also call myself a “datum”, the singular form of data, literally one piece of information to solve a bigger puzzle. Honestly, I don’t need 10,000 years to become a datum. I am already a part of several datasets people use these days to study many things. So for that objective learner, I am already a datum. Now let me get to how I got the dumb part of it. I have had an interesting and eventful life so far...
If you had data of the entire world at your feet, would you still build a deep learning model on it?
Suppose you solved a crucial business problem by building a complex deep neural network model to generate predictions on a new set of business data. Deploying and maintaining this model costs your business a fortune, but the decision-makers consider it a small price to pay for a greater good. It is all well and good. But could you have optimized the cost by opting for a simple regression model instead of a deep neural network? Of course, you tried regression as a baseline model, and you went for a deep neural network only after you noticed that it outperformed the regression model by miles. But if you had to suddenly build this model again with an enormous volume of training data, would you notice any difference in the performance metrics of the baseline and the complex models? Or, for that matter, would an increase in the volume of training data improve the performance of a baseline model and make it on par with a complex model? Two Microsoft researchers, Michele Banko and Eric ...