If you’re an absolute beginner to machine learning, I hope this post is useful to you.
In the recent years, AI, or machine learning has been talked about everywhere. It’s been a science fiction dream since decades ago and now humanity is finally making big progressses towards it. However, for the average joe, machine learning is misunderstood. Yes, machine learning can be very, very complicated, but to grasph the concepts of it can be quite simple.
To explain in one line: machine learning is about making a prediction based on data.
Do you recall middle school math class where you used a ruler to draw a “line-of-best-fit” into a set of points? The optimal “line-of-best-fit” would be the one that was closest to all the points.
With this line, and the premis that the new data follows the same distribution, you could predict where the new points will likely show up.
And this is exactly what machine learning is
There’s fancy mathematical ways to calculate the “line-of-best-fit” based on the dataset we have, and we “learn” what the “line-of-best-fit” is by drawing it out multiple times and trying to see which line fits best. This is exactly what Linear Regression in machine learning is.
There are other kinds too, notably, Logistic Regression and the newly arising Neural Network, but the principle is the same. Based on data we know, we draw the best line we can (via training) and try to predict results.
Now that you understand this, you also know that basic machine learning is far from generalized AI. Just because your program can make very good predictions, it does not necessarily mean it’s intelligent in the sense that most people think AI to be.