It began in the winter of 2019 when I came across the journal in Annual Reviews of Fluid Mechanics —Machine Learning for Fluid Mechanics. As a Physics enthusiast and hailing from a mechanical engineering background, fluid mechanics has always titillated me. My professional job elevated if not demanded the thirst for being prolific in the field of fluids. Hence, it was not capricious that I came across the above journal by Steven Brunton. However, the journal was recondite to my amateur mind. Terms like NN, CNN, RNN, LSTM, etc. were not greek to me but due to a lack of basic and clear understanding, their usage in the journal was not efficacious for me. I guess, it was at this moment I decided to take a walk to the yard which some describe as the Magnum Opus of modern technology — Artificial intelligence.
The copious resources on AI confounded the dawn of my journey. As an ardent book reader and after some browsing, I chose Artificial Intelligence-A Modern Approach. Undoubtedly, the book is laudable, but again my amateur brain was being pejorative to the concepts. After trying a few other texts based on Google's suggestions, I realized that I need a more pragmatic approach. At this moment, Coursera became my inn and it provided a great lodging for the basic understanding and difference of the terms AI, ML, DL, Data Science, Data Analysis, etc. I took a couple of professional courses offered by IBM- Professional Certificate in Data Science and Professional Certificate in AI Engineering. The former was a didactic course. I developed a lot of new skills and polished my previous skills in Python. There were moments of depression, little glory, excitement, lots of learning. Statistical Learning was in the air. I took another course on Coursera for Statistics with R. In fact, I read a couple of intermediate and advanced texts of Springer Texts in Statistics. It was all so beautiful until the realization of the clouds still hovering over that journal came into being. A dilatory feeling started to seep in. With lots of viscosity, I completed the Professional Course in AI engineering. That boosted me a little. It had 3 different courses on DL with Keras, TensorFlow, and PyTorch. But somehow PyTorch attracted me. For some reason feature of AutoGrad tallied with my love of mathematics. Also, a Capstone Project(with PyTorch)which I did understand and completed successfully after little turbulence. All over they were good courses and provided me lots of learning but there was still something not right. It was as if I was becoming more like a dilettante. Of course, it was not the courses but my choice. They were professional courses. They were not meant for a deeper dive and the same was conveyed in them. However, I needed something else, something which I got through an advertisement link. It was JovianML.
By now, I had a discrete knowledge of the concepts of DL. But the cohesive forces required to make it solid were missing due to a lack of clear understanding. In search of the cohesive forces, I have reached the inn of JovianML owned by Aakash NS. I did some digging and decided to take his course. As part of his course, there are weekly assignments and also an optional article writing on Medium for the same. So, I will try to post my current learnings and previous learnings every week. The stories will be of a mechanical engineer, physics enthusiast, non-programmer(please provide suggestions for an improved code), avid fluid mechanics guy, and a learner above all who went from Zero to GANs. And in doing so, maybe I will attain enlightenment and espouse Machine Learning for Fluid Mechanics. So, I guess now I can consider this moment as My First Encounter with Deep Learning.
In conclusion, AI is fascinating. It’s mesmerizing. It’s like those Patronus originating from the magic wands in Harry Potter. But it’s tantamount to know your requirements for the same. I learned it the hard way, maybe the reader can utilize it.
Your feedback will help an avid learner.
