1
00:00:00,000 --> 00:00:00,255


2
00:00:00,255 --> 00:00:01,410
VICKY: My name is Vicky.

3
00:00:01,410 --> 00:00:04,050
I'm from Canada.

4
00:00:04,050 --> 00:00:05,737
I think I got a little overeager.

5
00:00:05,737 --> 00:00:08,820
I asked two questions, and now I'm not
sure which one you're referring to.

6
00:00:08,820 --> 00:00:10,730
SPEAKER 1: I think you
had a math question.

7
00:00:10,730 --> 00:00:12,330
VICKY: OK, yes.

8
00:00:12,330 --> 00:00:15,190
So my question was a
little desperately worded.

9
00:00:15,190 --> 00:00:19,440
But it was, can you please promise
me that all of these linear algebra,

10
00:00:19,440 --> 00:00:23,595
and discrete math, and calculus classes
will come in handy in my future career?

11
00:00:23,595 --> 00:00:26,370
SPEAKER 1: Yeah, a common
sentiment for the group.

12
00:00:26,370 --> 00:00:29,130
This was asked essentially
in all capital letters, too.

13
00:00:29,130 --> 00:00:34,380
So let me turn to Brian, who's
studied these topics most recently

14
00:00:34,380 --> 00:00:39,690
and also, I think, can speak to just how
applicable they can be in some domains.

15
00:00:39,690 --> 00:00:42,487
BRIAN: Sure, well, thank
you for that question.

16
00:00:42,487 --> 00:00:45,570
I definitely felt the same way when I
was taking some of my linear algebra

17
00:00:45,570 --> 00:00:47,132
and calculus classes.

18
00:00:47,132 --> 00:00:48,840
I think ultimately
whether or not it will

19
00:00:48,840 --> 00:00:51,990
come in handy will depend in
part on what your career path is.

20
00:00:51,990 --> 00:00:54,570
But there are definitely
areas within computer science

21
00:00:54,570 --> 00:00:57,510
where calculus, linear algebra,
discrete math, where all of that

22
00:00:57,510 --> 00:01:01,590
becomes relevant, especially in the
world of artificial intelligence

23
00:01:01,590 --> 00:01:03,020
and machine learning nowadays.

24
00:01:03,020 --> 00:01:05,610
So as you think about
using calculus, you're

25
00:01:05,610 --> 00:01:07,980
doing it very abstractly
to do things like how

26
00:01:07,980 --> 00:01:10,173
do you find the maximum
point of a function.

27
00:01:10,173 --> 00:01:12,090
You could translate that
into machine learning

28
00:01:12,090 --> 00:01:15,480
setting to how do you maximize the
accuracy of your machine learning

29
00:01:15,480 --> 00:01:16,110
system.

30
00:01:16,110 --> 00:01:19,560
So a lot of the ideas that are taken
from calculus and linear algebra

31
00:01:19,560 --> 00:01:22,800
have a lot of application nowadays
in terms of modern machine

32
00:01:22,800 --> 00:01:24,510
learning research.

33
00:01:24,510 --> 00:01:27,870
Linear algebra, in particular, also
a lot of applications to graphics.

34
00:01:27,870 --> 00:01:30,370
So, some of the stuff that
Colton's just been talking about,

35
00:01:30,370 --> 00:01:31,920
about games and game development.

36
00:01:31,920 --> 00:01:35,460
A lot of graphics right now was
ultimately based on linear algebra.

37
00:01:35,460 --> 00:01:38,420
And a lot of what graphical
processing units are doing

38
00:01:38,420 --> 00:01:42,210
are just like pieces of hardware that
are designed to do these linear algebra

39
00:01:42,210 --> 00:01:44,860
operation very, very quickly.

40
00:01:44,860 --> 00:01:46,890
So definitely a lot of
applications, depending

41
00:01:46,890 --> 00:01:50,820
on where within computer science or
other domains you're interested in,

42
00:01:50,820 --> 00:01:53,586
but they do become
important that some point.

43
00:01:53,586 --> 00:01:55,430
SPEAKER 1: And I can
say, too, I felt some

44
00:01:55,430 --> 00:01:57,890
of the same frustrations
taking math classes in college

45
00:01:57,890 --> 00:01:59,310
and also in high school.

46
00:01:59,310 --> 00:02:02,960
And I think it's in part because of
how the classes I took were taught.

47
00:02:02,960 --> 00:02:05,900
Many of the math classes I
took were just so mechanical.

48
00:02:05,900 --> 00:02:08,900
Like it was just problem
after problem after problem.

49
00:02:08,900 --> 00:02:10,639
And I worry that a
lot of classes sort of

50
00:02:10,639 --> 00:02:13,350
lose sight of the forest
for the trees, so to speak,

51
00:02:13,350 --> 00:02:16,520
which is they focus on a lot
of the lower level details

52
00:02:16,520 --> 00:02:19,190
without appreciating that
what's really important

53
00:02:19,190 --> 00:02:21,950
is the higher level concepts
like what a derivative is

54
00:02:21,950 --> 00:02:24,560
or what an integral is
in the world of calculus.

55
00:02:24,560 --> 00:02:28,190
And frankly, in retrospect-- and I can
say this with some confidence 20 plus

56
00:02:28,190 --> 00:02:28,970
years later--

57
00:02:28,970 --> 00:02:32,930
I really didn't need to know 12
different ways to take a derivative

58
00:02:32,930 --> 00:02:36,200
or do an integral, especially
now when we have computers that

59
00:02:36,200 --> 00:02:37,940
can help with some of those processes.

60
00:02:37,940 --> 00:02:40,520
Hands down important to
understand the applicability

61
00:02:40,520 --> 00:02:43,100
of finding the min or the
max and what problems you

62
00:02:43,100 --> 00:02:45,860
can solve with those kinds of
techniques, let alone matrices

63
00:02:45,860 --> 00:02:48,110
and the like in the
world of linear algebra.

64
00:02:48,110 --> 00:02:51,620
But take some comfort in knowing that
even though courses you're taking now

65
00:02:51,620 --> 00:02:55,310
in math might be kind of belaboring
the point again and again,

66
00:02:55,310 --> 00:02:59,660
the ideas are useful even if you
start to forget some of the mechanics

67
00:02:59,660 --> 00:03:04,960
and don't get all of the answers right
when trying to do things by hand.

68
00:03:04,960 --> 00:03:06,000