Check out a free preview of the full A Practical Guide to Machine Learning with TensorFlow 2.0 & Keras course

The "Statistics & Machine Learning" Lesson is part of the full, A Practical Guide to Machine Learning with TensorFlow 2.0 & Keras course featured in this preview video. Here's what you'd learn in this lesson:

Vadim discusses the relationship between statistics and machine learning, and explains the concept of feeding. Feeding is when machine learning takes practical information and measurements, and finds distributions and corresponding values.


Transcript from the "Statistics & Machine Learning" Lesson

>> What is machine learning? I kinda start already showing you examples of machine learning, deep learning, but I haven't even explained what that actually is. So, for that, I will be using kinda by example. Show you how we can differentiate between statistics, machine learning, deep learning, and even AI.

So let's start with just statistics, what is statistics? And let's start with some practical problem, let's say I'm a shoemaker in 19th century. And have no idea about machine learning distributions and everything else. But I want to create really good shoes and they know I'm creating them. But I don't know what's the optimal size I should use to basically provide to my customers.

And let's say I want to massively produce them, right? And hope that it will satisfy the needs of my customers, so what I can do? I can just go to 100 random strangers, right? And ask them what shoe size do they have, let's say 8, 9, 10, 11, 12, 13, I don't know.

And for instance, some people say, 5 people say they have size 8. More people will say that they have size 9, and so on and so forth, so I can almost plot of this distribution of sizes, right? And statistics allows me to just get those 100 numbers and find the average, for instance, or also mean.

Is the same thing which is just a summation of all of those numbers divided by number of people participated in my survey. So, I just simply can say that, if I go with the mean size, which in my case probably going to be number 11, I will satisfy the majority, right?

And probably people with size 8 and 9, not gonna be satisfied with big shoes with size 11, and 12 and 13 will not fit. But still, at least big chunk of people will, well, get their shoes the right size. So statistics allows us to find some basic information about the distributions, right?

And then statistics kinda evaluated into machine learning, machine learning, the way how I see it, it mostly deals with the distributions. So back to our shoe size example, that looks almost like a normal distribution. If you not familiar with the normal distributions or Gaussian distributions, or bell curve distribution.

It's called this because it is kinda look like a bell curve. It is symmetric, I'm just drawing it, it's not that symmetrical, and it can be easily described by two numbers. The kinda mean value, where we have this peak and standard deviation, think about it like the width of the distribution in the middle.

So, machine learning, pretty much what it's doing all the time is trying to find the corresponding distribution. And find the values which can describe this distribution and this process called fitting. So machine learning is pretty good with taking a lot of practical observations, right? Different measurements and trying to find those distributions and corresponding values which can describe this distribution.

And people been doing that for, not centuries, well, I can probably say centuries. And the thing is, at this point you don't need a lot of computer. Majority of those operations are requiring just summation, division multiplication, but all of that can be done by hands.

Learn Straight from the Experts Who Shape the Modern Web

  • In-depth Courses
  • Industry Leading Experts
  • Learning Paths
  • Live Interactive Workshops
Get Unlimited Access Now