Indian Institute of Management Raipur and my story began in May'21. Master's was ending and minds were blocked. However, research ideas were still stimulating. I got desperate to begin something new. Flurried here and there but all in vain. Then Mihir suggested me to try at IIM Raipur. I was told that the professors at IIms are most easily approachable. Then why not give it a try? And indeed it was true. I contacted a professor and I got a response within a few hours. It was a conversation over mail and I got Hired! for conducting research under his mentorship. The prof wasn't much aware of my capabilities and he was taking a chance. (I have a slight idea that he was impressed by my profile.) He asked me to choose one of his three project ideas. I got to work on Bitcoin Price Prediction. The project was to implement an LSTM autoencoder regularized by False Nearest Neighbor loss. I knew LSTM very well back then. I knew autoencoders. But I had never heard about FNN. There was a paper on FNN by a Harvard professor which I never wanted to read (thinking Harvard papers won't ever get into my mind). So, I could implement the LSTM Autoencoder but not the FNN. Another trouble was, the reference codes were in R. I had never encountered an R code before. And now I had to learn R. It is said that R is easy but I never consider so at first about any programming language. (P.S.: I hated Programming.) Anyhow, I started implementing. I was getting closer somehow by repeatedly reading a single blog and the codes but not at all the Harvard guy's paper. Then came a day, R implementation was successful. Not to forget, the results. Those were dreadful! Then I was trying multiple ways to get around it. Some things worked out, some didn't. I wasn't satisfied. Then I thought to re-implement the code in Python. I thought it would be easier to optimize in a familiar language. Yes, it was. Everything worked out fine, except for the FNN code, again! I asked few of my friends for help but no good results. One of them asked me to go back and read the Harvard paper and then get back to him. You guessed it right, I didn't read it. My ideology is to first implement the things and then understand the underlying concepts. I might be wrong at this point, I admit. Somehow the code was fixed and honestly, there wasn't much to be done to fix it, which was realized later. My experiments smoothly began and I had good results. All of this happened within 10 days of time. I conveyed the results to my prof and began with the draft. I had to compile a 2000-word manuscript, second shortest of my life. And I felt I wasn't able to do it. I couldn't begin. I couldn't write a single word. And this was happenning after I had successfully written seven papers. All I was doing is procrastination. Its common among researchers. I had set a deadline in my mind to submit the manuscript by May 31. On 30th evening, I began to write. On June 1, I submitted the 1997-word draft. The professor reviewed it and asked me to make minor changes. (Probably he wants a 3000-word draft now.) And guess what? Its June 16th and I'm still procrastinating. I'll write the draft tomorrow, for sure!.
P.S. I have read that Harvard guy's paper.
And I
know R.