1
00:00:00,000 --> 00:00:04,700
[MUSIC]

2
00:00:04,700 --> 00:00:09,513
This is going to be a highly hands on
course, a little tough, hard work, but

3
00:00:09,513 --> 00:00:13,910
you're going to come out
really an expert on the topics.

4
00:00:13,910 --> 00:00:18,990
As usual we're going to keep on our core
philosophy, which is to always use case

5
00:00:18,990 --> 00:00:24,010
studies to motivate everything that we do
and you'll see that throughout the course.

6
00:00:24,010 --> 00:00:27,240
But there's also six things I
think about in every module and

7
00:00:27,240 --> 00:00:28,870
you'll see that happening.

8
00:00:28,870 --> 00:00:31,110
We'll talk about core concepts.

9
00:00:31,110 --> 00:00:35,100
Core concepts that go beyond
the techniques in that particular module.

10
00:00:35,100 --> 00:00:38,350
I've created visualizations for
every aspect of the course.

11
00:00:38,350 --> 00:00:41,280
You'll see new kinds of
visualizations that we hope

12
00:00:41,280 --> 00:00:44,020
will help you really grasp
the underlying ideas.

13
00:00:45,550 --> 00:00:49,100
We'll talk about actual
algorithms in detail

14
00:00:49,100 --> 00:00:51,830
that make it possible to build
these things in practice.

15
00:00:51,830 --> 00:00:54,380
So we'll really cover the algorithms but

16
00:00:54,380 --> 00:00:57,760
we won't cover everything
about classification.

17
00:00:57,760 --> 00:01:01,730
We'll narrow down the course to
cover techniques that are really

18
00:01:01,730 --> 00:01:04,370
impacting the world today,
that are actually practical.

19
00:01:04,370 --> 00:01:08,760
And we'll talk about what it takes to
make those techniques practical, and

20
00:01:08,760 --> 00:01:10,960
that's a big thing,
because you're going to implement them,

21
00:01:10,960 --> 00:01:14,340
you're going to implement them
from scratch in this course.

22
00:01:14,340 --> 00:01:17,330
All the algorithms that we're going to
talk about, you'll implement from scratch,

23
00:01:17,330 --> 00:01:20,810
and you'll be able to take them into the
real world, and implement them yourselves,

24
00:01:20,810 --> 00:01:24,260
if you're interested in them or
use them from a package.

25
00:01:24,260 --> 00:01:30,030
But, really understand where they're
coming from. For every module we cover in

26
00:01:30,030 --> 00:01:35,070
this course, we've created some additional
advanced topics in advanced areas.

27
00:01:35,070 --> 00:01:37,700
We're going to mark those as optional.

28
00:01:37,700 --> 00:01:40,910
For those interested, you'll be
able to dig in, for every module,

29
00:01:40,910 --> 00:01:44,280
new things you might want to learn
if you want to go into more depth.

30
00:01:44,280 --> 00:01:46,740
But those are optional
You don't need them.

31
00:01:46,740 --> 00:01:51,810
Even if you skip all of those you
still have a really fantastic handle

32
00:01:51,810 --> 00:01:56,190
on some of the most important techniques
that actually work in the real world

33
00:01:56,190 --> 00:01:57,940
to build classifiers.

34
00:01:57,940 --> 00:02:02,720
We're not going to cover everything but
we're going to cover in depth now for

35
00:02:02,720 --> 00:02:05,050
you to be able to really understand that.

36
00:02:05,050 --> 00:02:07,750
So we're going to focus on models
that make the most difference.

37
00:02:07,750 --> 00:02:13,040
So that's linear classifiers,
logistic regression, decision trees,

38
00:02:13,040 --> 00:02:14,490
and ensemble methods.

39
00:02:14,490 --> 00:02:19,780
And if you can understand those, you'll be
able to create the kinds of classifiers

40
00:02:19,780 --> 00:02:23,690
that you need to get amazing accuracy
on a wide range of real world problems.

41
00:02:24,890 --> 00:02:26,940
We're just going to talk
about fundamental algorithms,

42
00:02:26,940 --> 00:02:31,710
and you're going to implement them and
be very proficient at them.

43
00:02:31,710 --> 00:02:36,420
Gradient, which we also talked
about in the regression course.

44
00:02:36,420 --> 00:02:38,080
Stochastic gradient which is useful for

45
00:02:38,080 --> 00:02:42,110
scaling gradient methods and
classifiers to really massive problems.

46
00:02:43,280 --> 00:02:46,880
Recursive greedy algorithms that
are useful to learn decision trees.

47
00:02:46,880 --> 00:02:50,810
And Boosting, which is an amazing technique,
which is useful for

48
00:02:50,810 --> 00:02:53,570
running ensembles of classifiers.

49
00:02:53,570 --> 00:02:58,530
And we will discuss core machine learning
concepts which are useful way beyond

50
00:02:58,530 --> 00:03:00,160
the content of this course.

51
00:03:00,160 --> 00:03:01,470
Way beyond classifiers,

52
00:03:01,470 --> 00:03:06,180
so things like how to avoid overfitting,
how to deal with missing data.

53
00:03:07,300 --> 00:03:09,780
Precision-recall and online learning.

54
00:03:09,780 --> 00:03:12,814
It's going to be action-packed.

55
00:03:12,814 --> 00:03:16,969
[MUSIC]