Half One: The newbie part
After 6+ years of expertise with machine studying, each in analysis and trade, it’s very fascinating to see how far the sphere has gotten through the years. I nonetheless bear in mind sitting in seminar rooms, listening to lectures on all the pieces ML: deep studying, reinforcement studying, random forests, neural networks, pure language processing, …
One explicit NLP lecture stands out in my reminiscence, the place we mentioned the speedy developments within the area. We have been discussing vanilla consideration mechanisms the week earlier than and have been now taking a look at Transformer-based approaches. The nice tutor confirmed us graphs with the parameter counts of the fashions. We then seemed up the then-recent advances, and it turned clear: any determine is outdated inside a month. That was when my ML journey had barely began, and already the quantity of innovation that was revealed many times was insane.
Since then, my journey proceeded in a number of phases. From change with different ML folks, it seems that they’ve an analogous expertise: From a newbie to seasoned practitioner, the journey will be divided into the three following phases:
Part I: Newbie (~1 yr; this text)
Part II: Intermediate (~1 to three years; upcoming)
Part III: Superior (~5+ years; upcoming)