The World of Oreo-Eaters

A couple of weeks ago, I mentioned my oreo-eating personality type (twisting them open, eating the cream, then the cookie). I also confessed that this, statistically, means that I am intellectually curious, but that I can also be a cunning liar at times. I do not wish to kill my own credibility, but I am simply stating psychology statistics.

This week, I have come upon several revelations which have led me to wonder if the world is filled with people who eat oreos the way I do. I am not inverting the statement, but is this world truly filled with people who hide and mask the truth?

You may wonder what this has anything to do with deep learning (perhaps that certain matrices are not invertible) or professionalism (truth is a fundamental pillar of character), but I assure you that it is fully connected. This week, I have discovered how much we, students, have been lied to. Yes, we have been lied to, deceived, trapped in the cunning schemes of assumptions and misconceptions.

Although I greatly exaggerate, it is the truth. We may have been told false statements, but it is not considered lying because the exceptions were simply omitted. This week, I was able to conduct an interview with a Polymath Senior Research Scientist at NVIDIA, who was extremely generous to not only explain his career and field of work but offer invaluable advice regarding the journey of exploration and self-learning within Machine Learning. In our conversation, one of several points that struck me as intriguing was our perception of Academia.

I was being warned against the stagnation culture of the research community, so that I may pay more attention next time I read a research paper, lest I fall into the clutches of “peer-reviewed” and “academic journal” tags. With the sheer challenging aspect of making improvements to state-of-the-art technology, and potential pressures from third parties responsible for funding, perhaps, papers are churned out of the research community, without making significant-enough progress. According to what I understood from our discussion, tweaking learning rates to produce better results is not necessarily a revolutionary discovery. This means that there are several published papers out there, but not all of them may hold real improvements.

Why do I claim we, students, have been lied to? I believe we have been told, very specifically, that academic journals, peer-reviewed academic journals, are excellent sources of credible information and they can be trusted. This is true, no doubt, except for when it is not. Academic journals are supposed to be trusted, because schools and the academic community tells us so, because we trust our schools. True, research papers are credible, but the exceptions include papers churned out of the stagnation pressures. Apparently, there are times when papers may include equations that are incorrect, and remain uncorrected even after being informed of the error. In such a world, how can we trust anything?

Clearly, we have been lied to. However, I did mention that it is not lying since the exceptions were simply omitted. The credibility of a paper can be evaluated in several ways, the number of citations being one such measure. Being skeptical is another way, one advised by the research scientist I spoke to. Being careful and not skipping over details is crucial to not only understanding the problem and their solution, but it also helps increase one’s own ability to think critically.

Perhaps the world does eat oreos the way I do, but that does not mean we, students, have been lied to. We simply have had the “truth”, which is really a culmination of truths from different perspectives and viewpoints, revealed to us slowly, over time. Therefore, it is not lying if the world has not finished speaking, yet. Learning is progression, thus it suffices to say that learning consists of baby steps taken one at a time.

Along with taking small, specifically defined steps, direction is equally important. As vectors have magnitude and direction, learning, I have realized, requires specific steps taken in a specific direction. Going in random directions might lead to spirals and circles, which would be counterproductive. In the interview, I was advised to not go off in random directions, running after tools instead of concepts.

Although this was emphasizing what I have been told several times, I wondered how much I was or was not following this philosophy. I could claim that I know exactly what I want from a career perspective, but I would be lying to myself. I am using my time to explore, gather more data to make a more informed decision, hoping that I will be able to choose one path with conviction. After being told to not wander in random directions, I wondered how long I will have to wander before I find my direction. For a moment, I feared wandering forever, because that would make me a lost penguin in the desert under a cloudy sky, with no ability to fly nor find bearings. Then, I wondered how we could possibly differentiate between aimless wandering and purposeful exploration, both with vague destinations?