0
1
00:00:00,720 --> 00:00:02,410
Welcome back.
1

2
00:00:02,490 --> 00:00:05,950
Another day, another machine learning module.
2

3
00:00:06,240 --> 00:00:11,710
In this lesson I want to take a brief minute to talk about what we're going to do in this section.
3

4
00:00:11,730 --> 00:00:17,020
So far we've covered one machine learning technique, namely linear regression.
4

5
00:00:17,250 --> 00:00:21,060
And this was with one variable only. And
5

6
00:00:21,090 --> 00:00:25,950
also, we worked through the fundamentals of the Python programming language.
6

7
00:00:26,130 --> 00:00:27,990
So where do we go from here?
7

8
00:00:27,990 --> 00:00:35,280
Well, before we dive deeper into the next projects and tackle more complex machine learning techniques,
8

9
00:00:35,820 --> 00:00:43,320
we've got to peek under the hood and kind of get an understanding of how machine learning actually works.
9

10
00:00:43,320 --> 00:00:47,160
I mean, how does the learning in machine learning actually happen?
10

11
00:00:47,160 --> 00:00:53,640
What are the steps involved and how is it that we can give our computer a whole bunch of data and get
11

12
00:00:53,640 --> 00:00:56,430
back these beautiful parameter estimates?
12

13
00:00:56,460 --> 00:00:59,400
How are these calculations actually made?
13

14
00:00:59,400 --> 00:01:05,610
And to answer these and other utterly fascinating questions, we're going to have to get our hands dirty
14

15
00:01:05,790 --> 00:01:14,270
with some mathematics and some algorithms. And this means getting comfortable and working through a couple
15

16
00:01:14,270 --> 00:01:16,240
of difficult concepts.
16

17
00:01:16,400 --> 00:01:21,800
And that includes this idea of optimization, mathematical optimization.
17

18
00:01:21,800 --> 00:01:27,230
So you've probably already guessed that this section is going to be a bit more technical than the previous
18

19
00:01:27,230 --> 00:01:28,370
modules.
19

20
00:01:28,520 --> 00:01:30,650
We're going to be doing some calculus.
20

21
00:01:30,680 --> 00:01:33,830
We're going to work with derivatives and partial derivatives.
21

22
00:01:33,830 --> 00:01:39,290
We're going to be doing some more advanced programming Python too, because we're gonna be using loops and
22

23
00:01:39,290 --> 00:01:46,370
we're going to write our own optimization algorithm from scratch. And the optimization algorithm in question
23

24
00:01:46,730 --> 00:01:49,760
is called Gradient Descent.
24

25
00:01:49,760 --> 00:01:56,240
And once we've done all that, we're gonna apply this algorithm to various problems and see how it behaves.
25

26
00:01:57,140 --> 00:02:03,470
And what I mean by that is - in what situations does the algorithm do well? And when does this algorithm
26

27
00:02:03,470 --> 00:02:05,540
fall flat on its face?
27

28
00:02:05,570 --> 00:02:08,520
What are the strengths and weaknesses of the approach?
28

29
00:02:08,540 --> 00:02:10,620
What is the algorithm sensitive to?
29

30
00:02:10,850 --> 00:02:18,140
And does our algorithm have an allergy to peanuts? And all this will help us build an understanding of
30

31
00:02:18,140 --> 00:02:21,510
how machine learning actually works from the ground up.
31

32
00:02:21,680 --> 00:02:27,380
And if we want to be able to understand the more complex machine learning techniques like neural networks
32

33
00:02:27,620 --> 00:02:32,750
and tackle the more difficult projects, we're going to have to build up an understanding of how these
33

34
00:02:32,750 --> 00:02:36,490
techniques work to arrive at a solution in the first place.
34

35
00:02:37,760 --> 00:02:40,010
And that about covers the intro.
35

36
00:02:40,010 --> 00:02:41,470
I'll see you in the next lesson.