How we teach computers to understand pictures | Fei Fei Li

1,159,394 views ใƒป 2015-03-23

TED


์•„๋ž˜ ์˜๋ฌธ์ž๋ง‰์„ ๋”๋ธ”ํด๋ฆญํ•˜์‹œ๋ฉด ์˜์ƒ์ด ์žฌ์ƒ๋ฉ๋‹ˆ๋‹ค.

๋ฒˆ์—ญ: Juhyeon Kim ๊ฒ€ํ† : Jihyeon J. Kim
00:14
Let me show you something.
0
14366
3738
์ด๊ฑธ ๋ณด์‹œ์ฃ .
00:18
(Video) Girl: Okay, that's a cat sitting in a bed.
1
18104
4156
(์˜์ƒ) ์†Œ๋…€: "๊ณ ์–‘์ด๊ฐ€ ์นจ๋Œ€์— ์•‰์•„ ์žˆ์Šต๋‹ˆ๋‹ค."
00:22
The boy is petting the elephant.
2
22260
4040
"์†Œ๋…„์ด ์ฝ”๋ผ๋ฆฌ๋ฅผ ์“ฐ๋‹ค๋“ฌ๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค."
00:26
Those are people that are going on an airplane.
3
26300
4354
"์‚ฌ๋žŒ๋“ค์ด ๋น„ํ–‰๊ธฐ์— ํƒ€๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค."
00:30
That's a big airplane.
4
30654
2810
"ํฐ ๋น„ํ–‰๊ธฐ์ž…๋‹ˆ๋‹ค."
00:33
Fei-Fei Li: This is a three-year-old child
5
33464
2206
์ด๊ฑด ์„ธ ์‚ด์งœ๋ฆฌ ์•„์ด๊ฐ€ ์‚ฌ์ง„์„ ๋ณด๊ณ  ์„ค๋ช…ํ•˜๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
00:35
describing what she sees in a series of photos.
6
35670
3679
00:39
She might still have a lot to learn about this world,
7
39349
2845
๊ทธ๋…€๋Š” ์•„์ง ์ด ์„ธ์ƒ์— ๋Œ€ํ•ด ๋ฐฐ์šธ ๊ฒƒ์ด ๋งŽ์ง€๋งŒ,
00:42
but she's already an expert at one very important task:
8
42194
4549
ํ•œ ๊ฐ€์ง€ ์ผ์—์„œ๋งŒํผ์€ ์ด๋ฏธ ์ „๋ฌธ๊ฐ€ ์ˆ˜์ค€์ž…๋‹ˆ๋‹ค.
00:46
to make sense of what she sees.
9
46743
2846
๋ณธ ๊ฒƒ์„ ์ดํ•ดํ•˜๋Š” ์ผ์ด์ฃ .
00:50
Our society is more technologically advanced than ever.
10
50229
4226
์šฐ๋ฆฌ ์‚ฌํšŒ๋Š” ๊ทธ ์–ด๋Š ๋•Œ๋ณด๋‹ค ๊ธฐ์ˆ ์ ์œผ๋กœ ์ง„๋ณดํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
00:54
We send people to the moon, we make phones that talk to us
11
54455
3629
์šฐ๋ฆฌ๋Š” ๋‹ฌ์— ์‚ฌ๋žŒ์„ ๋ณด๋‚ด๊ณ , ๋ง์„ ํ•˜๋Š” ์ „ํ™”๋ฅผ ๋งŒ๋“ค๊ฑฐ๋‚˜
00:58
or customize radio stations that can play only music we like.
12
58084
4946
์ข‹์•„ํ•˜๋Š” ๊ณก๋งŒ ๋ฐฉ์†กํ•˜๋Š” ๋งž์ถคํ˜• ๋ผ๋””์˜ค๋ฅผ ๋งŒ๋“ญ๋‹ˆ๋‹ค.
01:03
Yet, our most advanced machines and computers
13
63030
4055
๊ทธ๋Ÿฌ๋‚˜ ์ฒจ๋‹จ ๊ธฐ๊ณ„์™€ ์ปดํ“จํ„ฐ๋กœ๋„
01:07
still struggle at this task.
14
67085
2903
์• ๋ฅผ ๋จน๋Š” ์ผ์ด ์žˆ์Šต๋‹ˆ๋‹ค.
01:09
So I'm here today to give you a progress report
15
69988
3459
์ €๋Š” ์˜ค๋Š˜ ์ปดํ“จํ„ฐ ๋น„์ „ ์—ฐ๊ตฌ์˜
01:13
on the latest advances in our research in computer vision,
16
73447
4047
์ตœ์‹  ๋™ํ–ฅ์— ๋Œ€ํ•ด ๋งํ•˜๊ณ ์ž ํ•ฉ๋‹ˆ๋‹ค
01:17
one of the most frontier and potentially revolutionary
17
77494
4161
์ปดํ“จํ„ฐ ๊ณผํ•™์—์„œ ๊ฐ€์žฅ ์„ ๋„์ ์ด๊ณ  ํ˜๋ช…์ ์ธ ๊ธฐ์ˆ ์ด์ฃ .
01:21
technologies in computer science.
18
81655
3206
01:24
Yes, we have prototyped cars that can drive by themselves,
19
84861
4551
์Šค์Šค๋กœ ์šด์ „ํ•˜๋Š” ์ž๋™์ฐจ ์‹œํ—˜ํŒ์„ ๋งŒ๋“ค๋”๋ผ๋„
01:29
but without smart vision, they cannot really tell the difference
20
89412
3853
๋˜‘๋˜‘ํ•œ ์ธ์‹ ๋Šฅ๋ ฅ์ด ์—†๋‹ค๋ฉด
01:33
between a crumpled paper bag on the road, which can be run over,
21
93265
3970
๋„๋กœ ์œ„์— ์žˆ๋Š” ๊ฒƒ์ด ๋ฐŸ์•„๋„ ๋  ์ข…์ด ๋ด‰ํˆฌ์ธ์ง€
01:37
and a rock that size, which should be avoided.
22
97235
3340
ํ”ผํ•ด์•ผ ํ•  ๋Œ๋ฉ์ด์ธ์ง€ ๊ตฌ๋ถ„ํ•  ์ˆ˜ ์—†์Šต๋‹ˆ๋‹ค.
01:41
We have made fabulous megapixel cameras,
23
101415
3390
์ˆ˜๋ฐฑ๋งŒ ํ™”์†Œ์˜ ์—„์ฒญ๋‚œ ์นด๋ฉ”๋ผ๋ฅผ ๋งŒ๋“ค๋”๋ผ๋„
01:44
but we have not delivered sight to the blind.
24
104805
3135
์‹œ๊ฐ์žฅ์• ์ธ์˜ ๋ˆˆ์ด ๋˜์ง€๋Š” ๋ชปํ•ฉ๋‹ˆ๋‹ค.
01:48
Drones can fly over massive land,
25
108420
3305
๋ฌด์ธ๊ธฐ๊ฐ€ ๊ด‘ํ™œํ•œ ๋•…์„ ๋‚  ์ˆ˜ ์žˆ์–ด๋„
01:51
but don't have enough vision technology
26
111725
2134
์ปดํ“จํ„ฐ ๋น„์ „ ๊ธฐ์ˆ ์ด ์—†์œผ๋ฉด
01:53
to help us to track the changes of the rainforests.
27
113859
3461
์—ด๋Œ€ ์šฐ๋ฆผ์˜ ๋ณ€ํ™”๋ฅผ ์ถ”์ ํ•˜์ง€ ๋ชปํ•ฉ๋‹ˆ๋‹ค.
01:57
Security cameras are everywhere,
28
117320
2950
๊ฐ์‹œ ์นด๋ฉ”๋ผ๊ฐ€ ๋„์ฒ˜์— ์žˆ์–ด๋„
02:00
but they do not alert us when a child is drowning in a swimming pool.
29
120270
5067
์ˆ˜์˜์žฅ์—์„œ ๋ฌผ์— ๋น ์ง„ ์•„์ด๋ฅผ ๋ณด๊ณ  ์šฐ๋ฆฌ์—๊ฒŒ ๊ฒฝ๊ณ ํ•ด ์ฃผ์ง€๋Š” ์•Š์Šต๋‹ˆ๋‹ค.
02:06
Photos and videos are becoming an integral part of global life.
30
126167
5595
์‚ฌ์ง„๊ณผ ๋น„๋””์˜ค๋Š” ์ง€๊ตฌ ์ƒํ™œ์˜ ๋ถˆ๊ฐ€๊ฒฐํ•œ ๋ถ€๋ถ„์ด ๋˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
02:11
They're being generated at a pace that's far beyond what any human,
31
131762
4087
์–ด๋–ค ๊ฐœ์ธ์ด๋‚˜ ๋‹จ์ฒด๊ฐ€ ๋‹ค ๋ณผ ์ˆ˜ ์—†์„ ๋ถ„๋Ÿ‰์˜
02:15
or teams of humans, could hope to view,
32
135849
2783
์˜์ƒ์ด ๋งŒ๋“ค์–ด์ง€๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
02:18
and you and I are contributing to that at this TED.
33
138632
3921
์—ฌ๊ธฐ TED๋„ ์ผ์กฐํ•˜๊ณ  ์žˆ์ง€์š”.
02:22
Yet our most advanced software is still struggling at understanding
34
142553
5232
๊ทธ๋Ÿฌ๋‚˜ ๊ฐ€์žฅ ์ง„๋ณดํ•œ ์†Œํ”„ํŠธ์›จ์–ด๋„ ์•„์ง๊นŒ์ง€๋Š”
02:27
and managing this enormous content.
35
147785
3876
์ด ๋ฐฉ๋Œ€ํ•œ ์˜์ƒ์„ ์ดํ•ดํ•˜๊ณ  ๊ด€๋ฆฌํ•˜๋Š”๋ฐ ์• ๋ฅผ ๋จน๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
02:31
So in other words, collectively as a society,
36
151661
5272
๋‹ฌ๋ฆฌ ๋งํ•˜์ž๋ฉด ์‚ฌํšŒ ์ „์ฒด์ ์œผ๋กœ
02:36
we're very much blind,
37
156933
1746
์šฐ๋ฆฌ๋Š” ์žฅ๋‹˜๊ณผ ๊ฐ™์Šต๋‹ˆ๋‹ค.
02:38
because our smartest machines are still blind.
38
158679
3387
์šฐ๋ฆฌ์˜ ๊ฐ€์žฅ ๋˜‘๋˜‘ํ•œ ๊ธฐ๊ณ„๊ฐ€ ์•„์ง๊นŒ์ง€ ์žฅ๋‹˜์ด๋‹ˆ๊นŒ์š”.
02:43
"Why is this so hard?" you may ask.
39
163526
2926
"๊ทธ๊ฒŒ ์™œ ์–ด๋ ต์ง€?" ํ•˜๊ณ  ๋ฌผ์œผ์‹ค ์ˆ˜ ์žˆ์–ด์š”.
02:46
Cameras can take pictures like this one
40
166452
2693
์นด๋ฉ”๋ผ๋Š” ์ด๋Ÿฐ ์‚ฌ์ง„์„ ์ฐ์„ ์ˆ˜ ์žˆ๊ณ 
02:49
by converting lights into a two-dimensional array of numbers
41
169145
3994
๋น›์„ ์ˆซ์ž์˜ 2์ฐจ์› ๋ฐฐ์—ด์ธ
02:53
known as pixels,
42
173139
1650
ํ”ฝ์…€๋กœ ๋ณ€ํ™˜ํ•  ์ˆ˜ ์žˆ์ง€๋งŒ,
02:54
but these are just lifeless numbers.
43
174789
2251
์ด๋Š” ๊ทธ์ € ์ฃฝ์€ ์ˆซ์ž์ผ ๋ฟ์ž…๋‹ˆ๋‹ค.
02:57
They do not carry meaning in themselves.
44
177040
3111
๊ทธ ์ž์ฒด์— ์˜๋ฏธ๋Š” ์—†์Šต๋‹ˆ๋‹ค.
03:00
Just like to hear is not the same as to listen,
45
180151
4343
'๋“ค๋ฆฌ๋Š”' ๊ฒƒ๊ณผ '๋“ฃ๋Š”' ๊ฒƒ์ด ๋˜‘๊ฐ™์ง€ ์•Š๋“ฏ์ด
03:04
to take pictures is not the same as to see,
46
184494
4040
์‚ฌ์ง„์„ '์ฐ๋Š”' ๊ฒƒ๊ณผ '๋ณด๋Š”' ๊ฒƒ์€ ๋˜‘๊ฐ™์ง€ ์•Š์Šต๋‹ˆ๋‹ค.
03:08
and by seeing, we really mean understanding.
47
188534
3829
'๋ณธ๋‹ค'๋Š” ๋ง์—๋Š” '์ดํ•ดํ•œ๋‹ค'๋Š” ๋œป์ด ์žˆ์Šต๋‹ˆ๋‹ค.
03:13
In fact, it took Mother Nature 540 million years of hard work
48
193293
6177
์‚ฌ์‹ค ์ž์—ฐ์€ 5์–ต 4์ฒœ๋งŒ๋…„์— ๊ฑธ์ณ
03:19
to do this task,
49
199470
1973
์ด ์ž‘์—…์„ ํ–ˆ๋Š”๋ฐ์š”.
03:21
and much of that effort
50
201443
1881
๊ทธ ๋…ธ๋ ฅ์˜ ๋Œ€๋ถ€๋ถ„์€
03:23
went into developing the visual processing apparatus of our brains,
51
203324
5271
์šฐ๋ฆฌ ๋‡Œ์˜ ์‹œ๊ฐ์ฒ˜๋ฆฌ๋Šฅ๋ ฅ์„ ๋ฐœ๋‹ฌ์‹œํ‚ค๋Š”๋ฐ ์†Œ์š”๋˜์—ˆ๊ณ 
03:28
not the eyes themselves.
52
208595
2647
๋ˆˆ์„ ๋งŒ๋“œ๋Š”๋ฐ ์†Œ์š”๋˜์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค.
03:31
So vision begins with the eyes,
53
211242
2747
์‹œ๊ฐํ˜„์ƒ์€ ๋ˆˆ์—์„œ ์‹œ์ž‘๋˜์ง€๋งŒ
03:33
but it truly takes place in the brain.
54
213989
3518
์‚ฌ์‹ค์ƒ ๋‚˜ํƒ€๋‚˜๋Š” ๊ณณ์€ ๋‡Œ ์•ˆ์ชฝ์ด์ฃ .
03:38
So for 15 years now, starting from my Ph.D. at Caltech
55
218287
5060
์ €๋Š” ์ตœ๊ทผ 15๋…„๊ฐ„ ์บ˜๋ฆฌํฌ๋‹ˆ์•„ ๊ณต๋Œ€ ๋ฐ•์‚ฌ ๊ณผ์ •์—์„œ๋ถ€ํ„ฐ
03:43
and then leading Stanford's Vision Lab,
56
223347
2926
์Šคํƒ ํฌ๋“œ๋Œ€ ์ปดํ“จํ„ฐ ๋น„์ „ ์—ฐ๊ตฌ์‹ค์„ ์ด๋Œ๊ธฐ๊นŒ์ง€
03:46
I've been working with my mentors, collaborators and students
57
226273
4396
์ง€๋„๊ต์ˆ˜, ๊ณต๋™์—ฐ๊ตฌ์ž, ํ•™์ƒ๋“ค๊ณผ ํ•จ๊ป˜
03:50
to teach computers to see.
58
230669
2889
์ปดํ“จํ„ฐ์—๊ฒŒ '๋ณด๋Š” ๋ฒ•'์„ ๊ฐ€๋ฅด์ณ์™”์Šต๋‹ˆ๋‹ค.
03:54
Our research field is called computer vision and machine learning.
59
234658
3294
์ €ํฌ ์—ฐ๊ตฌ ๋ถ„์•ผ๋ฅผ ์ปดํ“จํ„ฐ ๋น„์ „๊ณผ ๊ธฐ๊ณ„ ํ•™์Šต์ด๋ผ๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
03:57
It's part of the general field of artificial intelligence.
60
237952
3878
์ธ๊ณต์ง€๋Šฅ ์ผ๋ฐ˜ ๋ถ„์•ผ์— ์†ํ•˜์ฃ .
04:03
So ultimately, we want to teach the machines to see just like we do:
61
243000
5493
๊ถ๊ทน์ ์œผ๋กœ ์šฐ๋ฆฌ๋Š” ๊ธฐ๊ณ„๊ฐ€ ์ธ๊ฐ„์ฒ˜๋Ÿผ ๋ณผ ์ˆ˜ ์žˆ๊ฒŒ ํ•˜๋ ค๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
04:08
naming objects, identifying people, inferring 3D geometry of things,
62
248493
5387
๋ฌผ์ฒด์™€ ์‚ฌ๋žŒ์„ ์‹๋ณ„ํ•˜๊ณ , 3์ฐจ์› ๊ธฐํ•˜๊ตฌ์กฐ๋ฅผ ์ถ”์ธกํ•˜๊ณ ,
04:13
understanding relations, emotions, actions and intentions.
63
253880
5688
๊ด€๊ณ„, ๊ฐ์ •, ํ–‰๋™๊ณผ ์˜๋„๋ฅผ ์ดํ•ดํ•˜๊ฒŒ ํ•˜๋Š” ๊ฒ๋‹ˆ๋‹ค.
04:19
You and I weave together entire stories of people, places and things
64
259568
6153
์—ฌ๋Ÿฌ๋ถ„๊ณผ ์ €๋Š” ํ•œ๋ฒˆ ๋ณด๊ธฐ๋งŒ ํ•ด๋„
04:25
the moment we lay our gaze on them.
65
265721
2164
์‚ฌ๋žŒ, ์žฅ์†Œ, ์‚ฌ๋ฌผ๋กœ ์ด์•ผ๊ธฐ๋ฅผ ์—ฎ์–ด๋‚ผ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
04:28
The first step towards this goal is to teach a computer to see objects,
66
268955
5583
์ด๋Ÿฐ ๋ชฉํ‘œ๋ฅผ ํ–ฅํ•œ ์ฒซ๊ฑธ์Œ์ด ์ปดํ“จํ„ฐ๋ฅผ ๊ฐ€๋ฅด์ณ
04:34
the building block of the visual world.
67
274538
3368
์‚ฌ๋ฌผ, ์‹œ๊ฐ ์„ธ๊ณ„์˜ ๊ตฌ์„ฑ์š”์†Œ๋ฅผ ๋ณด๊ฒŒ ํ•˜๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
04:37
In its simplest terms, imagine this teaching process
68
277906
4434
๊ฐ„๋‹จํžˆ ๋งํ•ด, ์ด๋Ÿฐ ํ•™์Šต ๊ณผ์ •์„ ์ƒ์ƒํ•ด๋ณด์„ธ์š”.
04:42
as showing the computers some training images
69
282340
2995
์ปดํ“จํ„ฐ์— ํŠน์ • ์‚ฌ๋ฌผ์˜ ํ›ˆ๋ จ์šฉ ์ด๋ฏธ์ง€๋ฅผ ๋ณด์—ฌ์ค๋‹ˆ๋‹ค.
04:45
of a particular object, let's say cats,
70
285335
3321
๊ณ ์–‘์ด๋ผ๊ณ  ํ•ด๋ณด์ฃ .
04:48
and designing a model that learns from these training images.
71
288656
4737
๊ทธ๋ฆฌ๊ณ  ๊ทธ ํ›ˆ๋ จ์šฉ ์ด๋ฏธ์ง€๋กœ ํ•™์Šตํ•˜๋Š” ๋ชจ๋ธ์„ ์„ค๊ณ„ํ•ฉ๋‹ˆ๋‹ค.
04:53
How hard can this be?
72
293393
2044
๊ฐ„๋‹จํ•˜๊ฒŒ ๋“ค๋ฆฌ๋Š”๋ฐ์š”. ์–ผ๋งˆ๋‚˜ ์–ด๋ ค์šธ ์ˆ˜ ์žˆ์„๊นŒ์š”?
04:55
After all, a cat is just a collection of shapes and colors,
73
295437
4052
๊ณ ์–‘์ด๋Š” ๋ชจ์–‘๊ณผ ์ƒ‰๊น”์˜ ์ง‘ํ•ฉ์ด๊ณ ,
04:59
and this is what we did in the early days of object modeling.
74
299489
4086
์ด๊ฒƒ์ด ์šฐ๋ฆฌ๊ฐ€ ์ดˆ์ฐฝ๊ธฐ ๊ฐ์ฒด ๋ชจ๋ธ๋ง์œผ๋กœ ํ•œ ์ผ์ด์ฃ .
05:03
We'd tell the computer algorithm in a mathematical language
75
303575
3622
์šฐ๋ฆฌ๋Š” ์ปดํ“จํ„ฐ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ˆ˜ํ•™์  ์–ธ์–ด๋กœ ํ‘œํ˜„ํ•ฉ๋‹ˆ๋‹ค.
05:07
that a cat has a round face, a chubby body,
76
307197
3343
๊ณ ์–‘์ด๋Š” ๋‘ฅ๊ทผ ์–ผ๊ตด๊ณผ ํ†ตํ†ตํ•œ ๋ชธ,
05:10
two pointy ears, and a long tail,
77
310540
2299
๋‘ ๊ฐœ์˜ ๋พฐ์กฑํ•œ ๊ท€, ๊ธด ๊ผฌ๋ฆฌ๊ฐ€ ์žˆ๋‹ค๊ณ  ๊ฐ€๋ฅด์นฉ๋‹ˆ๋‹ค.
05:12
and that looked all fine.
78
312839
1410
๋‹ค ๊ดœ์ฐฎ์•„ ๋ณด์˜€์Šต๋‹ˆ๋‹ค.
05:14
But what about this cat?
79
314859
2113
๊ทธ๋Ÿฐ๋ฐ ์ด ๊ณ ์–‘์ด๋Š” ์–ด๋–จ๊นŒ์š”?
05:16
(Laughter)
80
316972
1091
(์›ƒ์Œ)
05:18
It's all curled up.
81
318063
1626
๋ชธ์„ ๋ง๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
05:19
Now you have to add another shape and viewpoint to the object model.
82
319689
4719
์ด์ œ ๊ฐ์ฒด ๋ชจ๋ธ์— ๋‹ค๋ฅธ ๋ชจ์–‘๊ณผ ๊ด€์ ์„ ์ถ”๊ฐ€ํ•ฉ๋‹ˆ๋‹ค.
05:24
But what if cats are hidden?
83
324408
1715
๊ทธ๋Ÿฐ๋ฐ ๋งŒ์•ฝ ๊ณ ์–‘์ด๊ฐ€ ์ˆจ์–ด ์žˆ์œผ๋ฉด์š”?
05:27
What about these silly cats?
84
327143
2219
์ด๋Ÿฐ ์›ƒ๊ธฐ๋Š” ๊ณ ์–‘์ด๋“ค์€์š”?
05:31
Now you get my point.
85
331112
2417
์ด์ œ ์ œ ๋ง์„ ์•„์‹œ๊ฒ ์ฃ .
05:33
Even something as simple as a household pet
86
333529
3367
์ง‘์•ˆ์˜ ์• ์™„๋™๋ฌผ์ฒ˜๋Ÿผ ๋‹จ์ˆœํ•œ ์‚ฌ๋ฌผ์กฐ์ฐจ
05:36
can present an infinite number of variations to the object model,
87
336896
4504
๊ฐ์ฒด ๋ชจ๋ธ์— ๋ฌดํ•œํ•œ ๋ณ€ํ˜•์ด ์กด์žฌํ•  ์ˆ˜ ์žˆ๊ณ ,
05:41
and that's just one object.
88
341400
2233
๊ทธ๊ฒŒ ํ•œ ๊ฐœ์˜ ๊ฐ์ฒด์ผ ๋ฟ์ด์ฃ .
05:44
So about eight years ago,
89
344573
2492
์•ฝ 8๋…„ ์ „
05:47
a very simple and profound observation changed my thinking.
90
347065
5030
๋‹จ์ˆœํ•˜๊ณ ๋„ ๊นŠ์€ ๊ด€์ฐฐ์ด ์ œ ์ƒ๊ฐ์„ ๋ฐ”๊พธ์—ˆ์Šต๋‹ˆ๋‹ค.
05:53
No one tells a child how to see,
91
353425
2685
์•„์ด์—๊ฒŒ ๋ณด๋Š” ๋ฒ•์„ ๊ฐ€๋ฅด์น  ์ˆœ ์—†์ฃ .
05:56
especially in the early years.
92
356110
2261
ํŠนํžˆ ์–ด๋ฆฐ ์‹œ์ ˆ์— ๋ง์ด์ฃ .
05:58
They learn this through real-world experiences and examples.
93
358371
5000
์•„์ด๋“ค์€ ํ˜„์‹ค์„ธ๊ณ„์˜ ๊ฒฝํ—˜๊ณผ ์‚ฌ๋ก€๋กœ ๋ณด๋Š” ๋ฒ•์„ ๋ฐฐ์›๋‹ˆ๋‹ค.
06:03
If you consider a child's eyes
94
363371
2740
๋งŒ์•ฝ ์•„์ด์˜ ๋ˆˆ์„
06:06
as a pair of biological cameras,
95
366111
2554
์ƒ๋ฌผํ•™์  ์นด๋ฉ”๋ผ ํ•œ์Œ์ด๋ผ ์น˜๋ฉด
06:08
they take one picture about every 200 milliseconds,
96
368665
4180
200๋ฐ€๋ฆฌ์ดˆ๋งˆ๋‹ค ํ•œ ์žฅ์”ฉ ์‚ฌ์ง„์„ ์ฐ๋Š” ์…ˆ์ด์ฃ .
06:12
the average time an eye movement is made.
97
372845
3134
๋ˆˆ์ด ์›€์ง์ด๋Š” ํ‰๊ท  ์‹œ๊ฐ„์ด์—์š”.
06:15
So by age three, a child would have seen hundreds of millions of pictures
98
375979
5550
์•„์ด๋Š” ์„ธ ์‚ด๊นŒ์ง€ ์ˆ˜์–ต์žฅ์˜ ํ˜„์‹ค์„ธ๊ณ„ ์‚ฌ์ง„์„ ๋ณด๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
06:21
of the real world.
99
381529
1834
06:23
That's a lot of training examples.
100
383363
2280
๋ฐฉ๋Œ€ํ•œ ์–‘์˜ ํ•™์Šต ์‚ฌ๋ก€์ฃ .
06:26
So instead of focusing solely on better and better algorithms,
101
386383
5989
๊ทธ๋ž˜์„œ ์ œ ์ƒ๊ฐ์—” ๋” ๋‚˜์€ ์•Œ๊ณ ๋ฆฌ์ฆ˜์—๋งŒ ์ง‘์ค‘ํ•˜๊ธฐ๋ณด๋‹ค,
06:32
my insight was to give the algorithms the kind of training data
102
392372
5272
์•Œ๊ณ ๋ฆฌ์ฆ˜์— ์ฃผ๋Š” ํ•™์Šต ๋ฐ์ดํ„ฐ๋ฅผ
06:37
that a child was given through experiences
103
397644
3319
์•„์ด๊ฐ€ ๊ฒฝํ—˜ํ•˜๋Š” ๊ฒƒ๊ณผ ๊ฐ™์ด ๋งŒ๋“ค์–ด์•ผ ํ–ˆ์Šต๋‹ˆ๋‹ค.
06:40
in both quantity and quality.
104
400963
3878
์–‘์ ์œผ๋กœ๋‚˜ ์งˆ์ ์œผ๋กœ ๋ง์ด์ฃ .
06:44
Once we know this,
105
404841
1858
์ด๊ฑธ ์•Œ๊ฒŒ ๋˜์ž,
06:46
we knew we needed to collect a data set
106
406699
2971
์šฐ๋ฆฌ๋Š” ์ด์ „๋ณด๋‹ค ํ›จ์”ฌ ๋งŽ์€ ๋ฐ์ดํ„ฐ๋ฅผ ๋ชจ์•„์•ผ ํ–ˆ์Šต๋‹ˆ๋‹ค.
06:49
that has far more images than we have ever had before,
107
409670
4459
06:54
perhaps thousands of times more,
108
414129
2577
๊ฑฐ์˜ ์ˆ˜์ฒœ๋ฐฐ์˜€์ฃ .
06:56
and together with Professor Kai Li at Princeton University,
109
416706
4111
๊ทธ๋ž˜์„œ ์ „ ํ”„๋ฆฐ์Šคํ„ด ๋Œ€ํ•™์˜ ์นด์ด ๋ฆฌ ๊ต์ˆ˜์™€ ํ•จ๊ป˜
07:00
we launched the ImageNet project in 2007.
110
420817
4752
2007๋…„ ์ด๋ฏธ์ง€๋„ท ํ”„๋กœ์ ํŠธ๋ฅผ ์‹œ์ž‘ํ–ˆ์Šต๋‹ˆ๋‹ค.
07:05
Luckily, we didn't have to mount a camera on our head
111
425569
3838
๋‹คํ–‰ํžˆ๋„ ์šฐ๋ฆฌ๋Š” ๋จธ๋ฆฌ์— ์นด๋ฉ”๋ผ๋ฅผ ๋งค๋‹ฌ๊ณ 
07:09
and wait for many years.
112
429407
1764
๋ช‡๋…„์”ฉ ๊ธฐ๋‹ค๋ฆด ํ•„์š”๋Š” ์—†์—ˆ์Šต๋‹ˆ๋‹ค.
07:11
We went to the Internet,
113
431171
1463
์ธํ„ฐ๋„ท์ด ์žˆ์—ˆ๊ฑฐ๋“ ์š”.
07:12
the biggest treasure trove of pictures that humans have ever created.
114
432634
4436
์ธ๋ฅ˜๊ฐ€ ๋งŒ๋“  ์ตœ๋Œ€์˜ ์‚ฌ์ง„ ์ฐฝ๊ณ ์ฃ .
07:17
We downloaded nearly a billion images
115
437070
3041
์šฐ๋ฆฌ๋Š” ๊ฑฐ์˜ 10์–ต์žฅ์˜ ์ด๋ฏธ์ง€๋ฅผ ๋‹ค์šด๋กœ๋“œํ–ˆ๊ณ 
07:20
and used crowdsourcing technology like the Amazon Mechanical Turk platform
116
440111
5880
์•„๋งˆ์กด MTurk ๊ฐ™์€ ํฌ๋ผ์šฐ๋“œ ์†Œ์‹ฑ ๊ธฐ์ˆ ์„ ์‚ฌ์šฉํ•ด
07:25
to help us to label these images.
117
445991
2339
์ด๋ฏธ์ง€์— ๋ผ๋ฒจ์„ ๋ถ™์˜€์Šต๋‹ˆ๋‹ค.
07:28
At its peak, ImageNet was one of the biggest employers
118
448330
4900
๊ฐ€์žฅ ์ตœ๊ณ ์น˜์—์„œ๋Š” ์ด๋ฏธ์ง€๋„ท์ด
07:33
of the Amazon Mechanical Turk workers:
119
453230
2996
์•„๋งˆ์กด MTurk ์ผ๊พผ๋“ค์˜ ์ตœ๋Œ€ ๊ณ ์šฉ์ฃผ์˜€์Šต๋‹ˆ๋‹ค.
07:36
together, almost 50,000 workers
120
456226
3854
5๋งŒ๋ช… ๊ฐ€๊นŒ์šด ์ž‘์—…์ž๊ฐ€
07:40
from 167 countries around the world
121
460080
4040
์„ธ๊ณ„ 167๊ฐœ๊ตญ์—์„œ
07:44
helped us to clean, sort and label
122
464120
3947
์•ฝ 10์–ต์žฅ์˜ ํ›„๋ณด ์ด๋ฏธ์ง€์˜
07:48
nearly a billion candidate images.
123
468067
3575
์ •๋ฆฌ ๋ถ„๋ฅ˜ ์ž‘์—…์„ ๋„์™”์Šต๋‹ˆ๋‹ค.
07:52
That was how much effort it took
124
472612
2653
์•„์ด์˜ ์„ฑ์žฅ ์ดˆ๊ธฐ์—
07:55
to capture even a fraction of the imagery
125
475265
3900
์ด๋ฏธ์ง€์˜ ์ผ๋ถ€๋ผ๋„ ์ˆ˜์ง‘ํ•˜๋Š”๋ฐ
07:59
a child's mind takes in in the early developmental years.
126
479165
4171
์–ผ๋งˆ๋‚˜ ๋งŽ์€ ๋…ธ๋ ฅ์ด ๋“œ๋Š”๊ฐ€ ํ•˜๋Š” ๊ฒƒ๊ณผ ๊ฐ™์•˜์ฃ .
08:04
In hindsight, this idea of using big data
127
484148
3902
์ง€๋‚˜๊ณ  ๋ณด๋‹ˆ, ์ปดํ“จํ„ฐ ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํ›ˆ๋ จ์—
08:08
to train computer algorithms may seem obvious now,
128
488050
4550
๋น…๋ฐ์ดํ„ฐ๋ฅผ ์‚ฌ์šฉํ•œ๋‹ค๋Š” ์•„์ด๋””์–ด๋Š” ์ด์ œ ํ™•์‹คํ•œ ๊ฒƒ ๊ฐ™์Šต๋‹ˆ๋‹ค๋งŒ,
08:12
but back in 2007, it was not so obvious.
129
492600
4110
2007๋…„ ๋‹น์‹œ์—๋Š” ๊ทธ๋ ‡์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค.
08:16
We were fairly alone on this journey for quite a while.
130
496710
3878
์šฐ๋ฆฌ ํ˜ผ์ž ์ด๋Ÿฐ ์ผ์„ ํ•œ ๊ฒŒ ๊ฝค ์˜ค๋ž˜ ๋์Šต๋‹ˆ๋‹ค.
08:20
Some very friendly colleagues advised me to do something more useful for my tenure,
131
500588
5003
์นœ์ ˆํ•œ ๋™๋ฃŒ๋Š” ์ข…์‹ ๊ต์ˆ˜๊ฐ€ ๋˜๋ ค๋ฉด ๋” ์œ ์šฉํ•œ ์ผ์„ ํ•˜๋ผ๊ณ  ์กฐ์–ธํ–ˆ๊ณ ,
08:25
and we were constantly struggling for research funding.
132
505591
4342
์šฐ๋ฆฌ๋Š” ๋Š˜ ์—ฐ๊ตฌ ์ž๊ธˆ ๋ฌธ์ œ์— ์‹œ๋‹ฌ๋ ธ์ฃ .
08:29
Once, I even joked to my graduate students
133
509933
2485
์ €๋Š” ์ด๋ฏธ์ง€๋„ท์˜ ์ž๊ธˆ ์กฐ๋‹ฌ์„ ์œ„ํ•ด ์„ธํƒ์†Œ๋ฅผ ๋‹ค์‹œ ์—ด์–ด์•ผ๊ฒ ๋‹ค๊ณ 
08:32
that I would just reopen my dry cleaner's shop to fund ImageNet.
134
512418
4063
๋Œ€ํ•™์›์ƒ๋“ค์—๊ฒŒ ๋†๋‹ด์„ ํ–ˆ์ฃ .
08:36
After all, that's how I funded my college years.
135
516481
4761
์ œ๊ฐ€ ๋Œ€ํ•™ ํ•™๋น„๋ฅผ ๋งˆ๋ จํ•œ ๋ฐฉ๋ฒ•์ด๊ฑฐ๋“ ์š”.
08:41
So we carried on.
136
521242
1856
์šฐ๋ฆฌ๋Š” ๊ณ„์† ์ง„ํ–‰ํ–ˆ์Šต๋‹ˆ๋‹ค.
08:43
In 2009, the ImageNet project delivered
137
523098
3715
2009๋…„์— ์ด๋ฏธ์ง€๋„ท ํ”„๋กœ์ ํŠธ๋Š”
08:46
a database of 15 million images
138
526813
4042
๊ฐ์ฒด์™€ ์‚ฌ๋ฌผ์„ 2๋งŒ2์ฒœ๊ฐœ ๋ฒ”์ฃผ๋กœ ๋ถ„๋ฅ˜ํ•œ
08:50
across 22,000 classes of objects and things
139
530855
4805
1์ฒœ5๋ฐฑ๋งŒ์žฅ ์ด๋ฏธ์ง€์˜ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค๋ฅผ ๋งŒ๋“ค์—ˆ๊ณ 
08:55
organized by everyday English words.
140
535660
3320
์ผ์ƒ์ ์ธ ์˜๋‹จ์–ด๋กœ ํ‘œํ˜„ํ–ˆ์Šต๋‹ˆ๋‹ค.
08:58
In both quantity and quality,
141
538980
2926
์–‘์ ์œผ๋กœ๋‚˜ ์งˆ์ ์œผ๋กœ๋‚˜
09:01
this was an unprecedented scale.
142
541906
2972
์ „๋ก€ ์—†๋Š” ๊ทœ๋ชจ์˜€์ฃ .
09:04
As an example, in the case of cats,
143
544878
3461
์˜ˆ๋ฅผ ๋“ค์–ด, ๊ณ ์–‘์ด์˜ ๊ฒฝ์šฐ
09:08
we have more than 62,000 cats
144
548339
2809
6๋งŒ 2์ฒœ์žฅ์˜ ์ด๋ฏธ์ง€๊ฐ€
09:11
of all kinds of looks and poses
145
551148
4110
๋‹ค์–‘ํ•œ ๋ชจ์–‘๊ณผ ์ž์„ธ,
09:15
and across all species of domestic and wild cats.
146
555258
5223
์ง‘๊ณ ์–‘์ด๋ถ€ํ„ฐ ๋“ค๊ณ ์–‘์ด๊นŒ์ง€ ๋ชจ๋“  ์ข…๋ฅ˜๋ฅผ ๋ง๋ผํ•ฉ๋‹ˆ๋‹ค.
09:20
We were thrilled to have put together ImageNet,
147
560481
3344
์šฐ๋ฆฌ๋Š” ์ด๋ฏธ์ง€๋„ท์„ ๋งŒ๋“  ๊ฒƒ์— ํฅ๋ถ„ํ–ˆ๊ณ 
09:23
and we wanted the whole research world to benefit from it,
148
563825
3738
๋ชจ๋“  ์—ฐ๊ตฌ์ž๋“ค๊ณผ ํ˜œํƒ์„ ๋‚˜๋ˆ„๊ณ ์ž ํ–ˆ์Šต๋‹ˆ๋‹ค.
09:27
so in the TED fashion, we opened up the entire data set
149
567563
4041
๊ทธ๋ž˜์„œ TED ๋ฐฉ์‹์œผ๋กœ ๋ชจ๋“  ๋ฐ์ดํ„ฐ๋ฅผ
09:31
to the worldwide research community for free.
150
571604
3592
์ „์„ธ๊ณ„์˜ ์—ฐ๊ตฌ์ž ์ปค๋ฎค๋‹ˆํ‹ฐ์— ๋ฌด๋ฃŒ๋กœ ๊ณต๊ฐœํ–ˆ์Šต๋‹ˆ๋‹ค.
09:36
(Applause)
151
576636
4000
(๋ฐ•์ˆ˜)
09:41
Now that we have the data to nourish our computer brain,
152
581416
4538
์ด์ œ ์šฐ๋ฆฌ๋Š” ์ปดํ“จํ„ฐ ๋‘๋‡Œ์— ์˜์–‘์„ ๊ณต๊ธ‰ํ•  ๋ฐ์ดํ„ฐ๊ฐ€ ์žˆ๊ณ ,
09:45
we're ready to come back to the algorithms themselves.
153
585954
3737
์•Œ๊ณ ๋ฆฌ์ฆ˜ ์ž์ฒด๋กœ ๋Œ์•„์˜ฌ ์ค€๋น„๊ฐ€ ๋˜์—ˆ์ฃ .
09:49
As it turned out, the wealth of information provided by ImageNet
154
589691
5178
๊ฒฐ๊ณผ์ ์œผ๋กœ ์ด๋ฏธ์ง€๋„ท์˜ ํ’๋ถ€ํ•œ ์ •๋ณด๋Š”
09:54
was a perfect match to a particular class of machine learning algorithms
155
594869
4806
๊ธฐ๊ณ„ ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŠน์ • ๋ถ„๋ฅ˜์— ๋”ฑ ๋“ค์–ด๋งž์•˜๋Š”๋ฐ,
09:59
called convolutional neural network,
156
599675
2415
์ด๋ฅผ ํ•ฉ์„ฑ๊ณฑ ์‹ ๊ฒฝ๋ง์ด๋ผ๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
10:02
pioneered by Kunihiko Fukushima, Geoff Hinton, and Yann LeCun
157
602090
5248
์ฟ ๋‹ˆํžˆ์ฝ” ํ›„์ฟ ์‹œ๋งˆ, ์ œํ”„๋ฆฌ ํžŒํŠผ, ์–‘ ๋ฃจ์บ‰์ด
10:07
back in the 1970s and '80s.
158
607338
3645
1970~80๋…„๋Œ€์— ๊ฐœ์ฒ™ํ•œ ์˜์—ญ์ด์ฃ .
10:10
Just like the brain consists of billions of highly connected neurons,
159
610983
5619
๋งˆ์น˜ ๋‡Œ๊ฐ€ ๊ณ ๋„๋กœ ์—ฐ๊ฒฐ๋œ ๋‰ด๋Ÿฐ ์ˆ˜์‹ญ์–ต๊ฐœ๋กœ ๊ตฌ์„ฑ๋œ ๊ฒƒ์ฒ˜๋Ÿผ
10:16
a basic operating unit in a neural network
160
616602
3854
์‹ ๊ฒฝ๋ง์˜ ๊ธฐ๋ณธ ๋‹จ์œ„๋Š”
10:20
is a neuron-like node.
161
620456
2415
๋‰ด๋Ÿฐ๊ณผ ๊ฐ™์€ ๋…ธ๋“œ์ž…๋‹ˆ๋‹ค.
10:22
It takes input from other nodes
162
622871
2554
๋‹ค๋ฅธ ๋…ธ๋“œ์—์„œ ์ž…๋ ฅ์„ ๋ฐ›๊ณ 
10:25
and sends output to others.
163
625425
2718
๋‹ค๋ฅธ ๋…ธ๋“œ๋กœ ์ถœ๋ ฅ์„ ๋ณด๋ƒ…๋‹ˆ๋‹ค.
10:28
Moreover, these hundreds of thousands or even millions of nodes
164
628143
4713
๊ฒŒ๋‹ค๊ฐ€ ์ด๋Ÿฐ ์ˆ˜์‹ญ๋งŒ, ์ˆ˜๋ฐฑ๋งŒ์˜ ๋…ธ๋“œ๋Š”
10:32
are organized in hierarchical layers,
165
632856
3227
๊ณ„์ธต ํ˜•ํƒœ๋กœ ์กฐ์งํ™”๋ฉ๋‹ˆ๋‹ค.
10:36
also similar to the brain.
166
636083
2554
๋‡Œ์™€ ๋งˆ์ฐฌ๊ฐ€์ง€์ฃ .
10:38
In a typical neural network we use to train our object recognition model,
167
638637
4783
์šฐ๋ฆฌ๊ฐ€ ์‚ฌ๋ฌผ ์ธ์‹ ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋ ค๊ณ  ์‚ฌ์šฉํ•œ ์ „ํ˜•์ ์ธ ์‹ ๊ฒฝ๋ง์—๋Š”
10:43
it has 24 million nodes,
168
643420
3181
2์ฒœ4๋ฐฑ๋งŒ์˜ ๋…ธ๋“œ,
10:46
140 million parameters,
169
646601
3297
1์–ต4์ฒœ๋งŒ์˜ ๋งค๊ฐœ๋ณ€์ˆ˜,
10:49
and 15 billion connections.
170
649898
2763
150์–ต์˜ ๊ฒฐํ•ฉ์ด ์กด์žฌํ•ฉ๋‹ˆ๋‹ค.
10:52
That's an enormous model.
171
652661
2415
์–ด๋งˆ์–ด๋งˆํ•œ ๋ชจ๋ธ์ด์ฃ .
10:55
Powered by the massive data from ImageNet
172
655076
3901
์ด๋ฏธ์ง€๋„ท์˜ ๋ฐฉ๋Œ€ํ•œ ๋ฐ์ดํ„ฐ์™€
10:58
and the modern CPUs and GPUs to train such a humongous model,
173
658977
5433
ํ˜„๋Œ€์˜ CPU์™€ GPU์— ํž˜์ž…์–ด
11:04
the convolutional neural network
174
664410
2369
ํ•ฉ์„ฑ๊ณฑ ์‹ ๊ฒฝ๋ง์€
11:06
blossomed in a way that no one expected.
175
666779
3436
์•„๋ฌด๋„ ์˜ˆ์ƒ์น˜ ๋ชปํ•œ ๋ฐฉ์‹์œผ๋กœ ๊ฝƒํ”ผ์—ˆ์Šต๋‹ˆ๋‹ค.
11:10
It became the winning architecture
176
670215
2508
์‚ฌ๋ฌผ์˜ ์ธ์‹์— ์žˆ์–ด
11:12
to generate exciting new results in object recognition.
177
672723
5340
ํฅ๋ฏธ๋กญ๊ณ ๋„ ์ƒˆ๋กœ์šด ๊ฒฐ๊ณผ๋ฅผ ๋‚ด๋Š” ์šฐ์ˆ˜ํ•œ ๊ตฌ์กฐ๊ฐ€ ๋˜์—ˆ์Šต๋‹ˆ๋‹ค.
11:18
This is a computer telling us
178
678063
2810
์ด ์ปดํ“จํ„ฐ๋Š” ์šฐ๋ฆฌ์—๊ฒŒ
11:20
this picture contains a cat
179
680873
2300
์ด ์‚ฌ์ง„์— ๊ณ ์–‘์ด๊ฐ€ ์žˆ๋Š”์ง€,
11:23
and where the cat is.
180
683173
1903
์–ด๋””์— ์žˆ๋Š”์ง€ ๋งํ•ด์ค๋‹ˆ๋‹ค.
11:25
Of course there are more things than cats,
181
685076
2112
๋ฌผ๋ก  ๊ณ ์–‘์ด ์ด์™ธ์˜ ๊ฒƒ๋„ ์ธ์‹ํ•  ์ˆ˜ ์žˆ๊ณ ,
11:27
so here's a computer algorithm telling us
182
687188
2438
์—ฌ๊ธฐ์„œ ์ปดํ“จํ„ฐ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ์‚ฌ์ง„ ์†์—
11:29
the picture contains a boy and a teddy bear;
183
689626
3274
์†Œ๋…„๊ณผ ํ…Œ๋”” ๋ฒ ์–ด๊ฐ€ ์žˆ๋‹ค๊ณ  ๋งํ•ด์ค๋‹ˆ๋‹ค.
11:32
a dog, a person, and a small kite in the background;
184
692900
4366
๊ฐœ, ์‚ฌ๋žŒ, ๋ฐฐ๊ฒฝ์— ์ž‘์€ ์—ฐ์ด ์žˆ์Šต๋‹ˆ๋‹ค.
11:37
or a picture of very busy things
185
697266
3135
๋˜๋Š” ๋งŽ์€ ๊ฒƒ์ด ์ฐํžŒ ์‚ฌ์ง„์—์„œ
11:40
like a man, a skateboard, railings, a lampost, and so on.
186
700401
4644
์‚ฌ๋žŒ, ์Šค์ผ€์ดํŠธ ๋ณด๋“œ, ๋‚œ๊ฐ„, ๊ฐ€๋กœ๋“ฑ ๊ฐ™์€ ๊ฒƒ์„ ๊ฐ€๋ ค๋ƒ…๋‹ˆ๋‹ค.
11:45
Sometimes, when the computer is not so confident about what it sees,
187
705045
5293
๋•Œ๋•Œ๋กœ ์ปดํ“จํ„ฐ๊ฐ€ ๋ณด๋Š” ๊ฒƒ์ด ๋ฌด์—‡์ธ์ง€ ํ™•์‹ ํ•˜์ง€ ๋ชปํ•  ๋•Œ๋Š”
11:51
we have taught it to be smart enough
188
711498
2276
์šฐ๋ฆฌ๋Š” ์ปดํ“จํ„ฐ๋ฅผ ๊ฐ€๋ฅด์ณ์„œ
11:53
to give us a safe answer instead of committing too much,
189
713774
3878
์–ต์ธก์„ ํ•˜๊ธฐ ๋ณด๋‹ค๋Š” ์•ˆ์ „ํ•œ ๋Œ€๋‹ต์„ ํ•˜๊ฒŒ ํ•ฉ๋‹ˆ๋‹ค.
11:57
just like we would do,
190
717652
2811
์‚ฌ๋žŒ๊ณผ ๋งˆ์ฐฌ๊ฐ€์ง€์ฃ .
12:00
but other times our computer algorithm is remarkable at telling us
191
720463
4666
๋ฐ˜๋ฉด ์ปดํ“จํ„ฐ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ๋†€๋ž๊ฒŒ๋„
12:05
what exactly the objects are,
192
725129
2253
์‚ฌ๋ฌผ์ด ์ •ํ™•ํžˆ ๋ฌด์—‡์ธ์ง€ ๋งํ•ด์ฃผ๊ธฐ๋„ ํ•ฉ๋‹ˆ๋‹ค.
12:07
like the make, model, year of the cars.
193
727382
3436
์ž๋™์ฐจ์˜ ์ฐจ์ข…, ๋ชจ๋ธ, ์—ฐ์‹ ๊ฐ™์€ ๊ฒƒ์ด์ฃ .
12:10
We applied this algorithm to millions of Google Street View images
194
730818
5386
์ˆ˜๋ฐฑ๊ฐœ ๋ฏธ๊ตญ ๋„์‹œ์—์„œ ์ฐ์€ ๊ตฌ๊ธ€ ์Šคํฌ๋ฆฌํŠธ ๋ทฐ ์ด๋ฏธ์ง€
12:16
across hundreds of American cities,
195
736204
3135
์ˆ˜๋ฐฑ๋งŒ์žฅ์— ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ ์šฉํ–ˆ๋”๋‹ˆ
12:19
and we have learned something really interesting:
196
739339
2926
ํฅ๋ฏธ๋กœ์šด ๊ฒƒ์„ ๋ฐœ๊ฒฌํ–ˆ์Šต๋‹ˆ๋‹ค.
12:22
first, it confirmed our common wisdom
197
742265
3320
๋จผ์ €, ์ผ๋ฐ˜์ ์œผ๋กœ ์˜ˆ์ƒํ•˜๋“ฏ์ด
12:25
that car prices correlate very well
198
745585
3290
์ž๋™์ฐจ ๊ฐ€๊ฒฉ์ด ๊ฐ€๊ณ„ ์ˆ˜์ž…๊ณผ
12:28
with household incomes.
199
748875
2345
๋งค์šฐ ๊ด€๋ จ์ด ์žˆ๋‹ค๋Š” ๊ฒƒ์ด์—ˆ์Šต๋‹ˆ๋‹ค.
12:31
But surprisingly, car prices also correlate well
200
751220
4527
ํ•˜์ง€๋งŒ ๋†€๋ž๊ฒŒ๋„, ์ž๋™์ฐจ ๊ฐ€๊ฒฉ์€
12:35
with crime rates in cities,
201
755747
2300
๋„์‹œ์˜ ๋ฒ”์ฃ„์œจ๊ณผ๋„ ๊ด€๋ จ์ด ์žˆ์—ˆ๊ณ ,
12:39
or voting patterns by zip codes.
202
759007
3963
๋„์‹œ๊ตฌ์—ญ๋ณ„ ํˆฌํ‘œ ๊ฒฝํ–ฅ๊ณผ๋„ ๊ด€๋ จ์ด ์žˆ์—ˆ์Šต๋‹ˆ๋‹ค.
12:44
So wait a minute. Is that it?
203
764060
2206
์ž ๊น๋งŒ์š”. ๊ทธ๋Ÿฐ๊ฐ€์š”?
12:46
Has the computer already matched or even surpassed human capabilities?
204
766266
5153
์ปดํ“จํ„ฐ๋Š” ์ด๋ฏธ ์ธ๊ฐ„์˜ ๋Šฅ๋ ฅ์„ ๋”ฐ๋ผ์žก๊ฑฐ๋‚˜ ์ถ”์›”ํ•œ ๊ฒƒ์ธ๊ฐ€์š”?
12:51
Not so fast.
205
771419
2138
๊ทธ๋ ‡์ง€๋Š” ์•Š์Šต๋‹ˆ๋‹ค.
12:53
So far, we have just taught the computer to see objects.
206
773557
4923
์ง€๊ธˆ๊นŒ์ง€ ์šฐ๋ฆฌ๋Š” ์ปดํ“จํ„ฐ์— ์‚ฌ๋ฌผ ์ธ์‹์„ ๊ฐ€๋ฅด์ณค์„ ๋ฟ์ด์—์š”.
12:58
This is like a small child learning to utter a few nouns.
207
778480
4644
๋งˆ์น˜ ์–ด๋ฆฐ ์•„์ด๊ฐ€ ๋ช…์‚ฌ ๋ช‡๊ฐœ๋ฅผ ๋ฐฐ์šด ๊ฒƒ๊ณผ ๊ฐ™์ฃ .
13:03
It's an incredible accomplishment,
208
783124
2670
์—„์ฒญ๋‚œ ์„ฑ๊ณผ์ด์ง€๋งŒ
13:05
but it's only the first step.
209
785794
2460
๊ทธ์ € ์ฒซ ๊ฑธ์Œ์— ๋ถˆ๊ณผํ•ฉ๋‹ˆ๋‹ค.
13:08
Soon, another developmental milestone will be hit,
210
788254
3762
๊ณง ๋‹ค์Œ ๊ฐœ๋ฐœ ๋ชฉํ‘œ์— ์ด๋ฅผ ๊ฒƒ์ด๊ณ ,
13:12
and children begin to communicate in sentences.
211
792016
3461
์–ด๋ฆฐ ์•„์ด๋Š” ๋ฌธ์žฅ์œผ๋กœ ์†Œํ†ต์„ ํ•˜๊ธฐ ์‹œ์ž‘ํ•  ๊ฒ๋‹ˆ๋‹ค.
13:15
So instead of saying this is a cat in the picture,
212
795477
4224
๊ทธ๋ž˜์„œ ์‚ฌ์ง„์„ ๋ณด๊ณ  '๊ณ ์–‘์ด์ž…๋‹ˆ๋‹ค' ํ•˜๋Š” ๋Œ€์‹ 
13:19
you already heard the little girl telling us this is a cat lying on a bed.
213
799701
5202
์—ฌ๋Ÿฌ๋ถ„์ด ์ด๋ฏธ ๋“ค์—ˆ๋“ฏ '๊ณ ์–‘์ด๊ฐ€ ์นจ๋Œ€์— ๋ˆ„์›Œ ์žˆ๋‹ค'๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
13:24
So to teach a computer to see a picture and generate sentences,
214
804903
5595
์ปดํ“จํ„ฐ๊ฐ€ ์‚ฌ์ง„์„ ๋ณด๊ณ  ๋ฌธ์žฅ์„ ๋งŒ๋“ค๊ฒŒ ๊ฐ€๋ฅด์น˜๋ ค๋ฉด,
13:30
the marriage between big data and machine learning algorithm
215
810498
3948
๋น… ๋ฐ์ดํ„ฐ์™€ ๊ธฐ๊ณ„ ํ•™์Šต ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ๊ฒฐํ•ฉ์ด
13:34
has to take another step.
216
814446
2275
๋˜ ํ•œ๋ฐœ์ง ๋‚˜์•„๊ฐ€์•ผ ํ•ฉ๋‹ˆ๋‹ค.
13:36
Now, the computer has to learn from both pictures
217
816721
4156
์ด์ œ ์ปดํ“จํ„ฐ๋Š” ์‚ฌ์ง„ ๋ฟ๋งŒ ์•„๋‹ˆ๋ผ
13:40
as well as natural language sentences
218
820877
2856
์‚ฌ๋žŒ์ด ๋งŒ๋“  ์ž์—ฐ ์–ธ์–ด ๋ฌธ์žฅ๋„
13:43
generated by humans.
219
823733
3322
๋ฐฐ์›Œ์•ผ ํ•ฉ๋‹ˆ๋‹ค.
13:47
Just like the brain integrates vision and language,
220
827055
3853
๋‡Œ๊ฐ€ ์‹œ๊ฐ๊ณผ ์–ธ์–ด๋ฅผ ๊ฒฐํ•ฉํ•˜๋“ฏ์ด,
13:50
we developed a model that connects parts of visual things
221
830908
5201
์šฐ๋ฆฌ๊ฐ€ ๊ฐœ๋ฐœํ•œ ๋ชจ๋ธ์€ ์ด๋ฏธ์ง€์˜ ๋‹จํŽธ๊ณผ ๊ฐ™์€
13:56
like visual snippets
222
836109
1904
์‹œ๊ฐ์  ์š”์†Œ๋ฅผ
13:58
with words and phrases in sentences.
223
838013
4203
๋ฌธ์žฅ ์† ๋‹จ์–ด๋‚˜ ๋ฌธ๊ตฌ์™€ ์—ฐ๊ฒฐํ•ฉ๋‹ˆ๋‹ค.
14:02
About four months ago,
224
842216
2763
์•ฝ 4๋‹ฌ ์ „
14:04
we finally tied all this together
225
844979
2647
์šฐ๋ฆฌ๋Š” ๋งˆ์นจ๋‚ด ์ด ๋ชจ๋‘๋ฅผ ์—ฎ์–ด
14:07
and produced one of the first computer vision models
226
847626
3784
์ตœ์ดˆ์˜ ์ปดํ“จํ„ฐ ๋น„์ „ ๋ชจ๋ธ ํ•˜๋‚˜๋ฅผ ๋งŒ๋“ค์—ˆ์Šต๋‹ˆ๋‹ค.
14:11
that is capable of generating a human-like sentence
227
851410
3994
์‚ฌ์ง„์„ ์ฒ˜์Œ ๋ณด์•˜์„๋•Œ ์‚ฌ๋žŒ๊ณผ ๊ฐ™์ด
14:15
when it sees a picture for the first time.
228
855404
3506
๋ฌธ์žฅ์„ ๋งŒ๋“ค์–ด๋‚ด๋Š” ๋ชจ๋ธ์ž…๋‹ˆ๋‹ค.
14:18
Now, I'm ready to show you what the computer says
229
858910
4644
์ด์ œ, ์—ฌ๋Ÿฌ๋ถ„๊ป˜ ์ปดํ“จํ„ฐ๊ฐ€ ์‚ฌ์ง„์„ ๋ณด๊ณ 
14:23
when it sees the picture
230
863554
1975
๋งํ•˜๋Š” ๊ฒƒ์„ ๋ณด์—ฌ๋“œ๋ฆฌ๊ฒ ์Šต๋‹ˆ๋‹ค.
14:25
that the little girl saw at the beginning of this talk.
231
865529
3830
์•ž์„œ ์–ด๋ฆฐ ์†Œ๋…€๊ฐ€ ๋ดค๋˜ ์‚ฌ์ง„์ž…๋‹ˆ๋‹ค.
(์ปดํ“จํ„ฐ) "๋‚จ์ž๊ฐ€ ์ฝ”๋ผ๋ฆฌ ์˜†์— ์„œ ์žˆ์Šต๋‹ˆ๋‹ค."
14:31
(Video) Computer: A man is standing next to an elephant.
232
871519
3344
14:36
A large airplane sitting on top of an airport runway.
233
876393
3634
"ํฐ ๋น„ํ–‰๊ธฐ๊ฐ€ ๊ณตํ•ญ ํ™œ์ฃผ๋กœ ๋์— ์žˆ์Šต๋‹ˆ๋‹ค."
14:41
FFL: Of course, we're still working hard to improve our algorithms,
234
881057
4212
๋ฌผ๋ก , ์šฐ๋ฆฌ๋Š” ์—ฌ์ „ํžˆ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ๊ฐœ๋Ÿ‰ํ•˜๋ ค๊ณ  ์ผํ•˜๊ณ  ์žˆ๊ณ 
14:45
and it still has a lot to learn.
235
885269
2596
๋ฐฐ์›Œ์•ผ ํ•  ๊ฒŒ ๋งŽ์Šต๋‹ˆ๋‹ค.
14:47
(Applause)
236
887865
2291
(๋ฐ•์ˆ˜)
14:51
And the computer still makes mistakes.
237
891556
3321
์ปดํ“จํ„ฐ๋Š” ์—ฌ์ „ํžˆ ์‹ค์ˆ˜๋ฅผ ์ €์ง€๋ฆ…๋‹ˆ๋‹ค.
14:54
(Video) Computer: A cat lying on a bed in a blanket.
238
894877
3391
(์ปดํ“จํ„ฐ) "๊ณ ์–‘์ด๊ฐ€ ์นจ๋Œ€ ์œ„ ์ด๋ถˆ ์•ˆ์— ์žˆ์Šต๋‹ˆ๋‹ค."
14:58
FFL: So of course, when it sees too many cats,
239
898268
2553
๊ณ ์–‘์ด๋ฅผ ๋„ˆ๋ฌด ๋งŽ์ด ๋ด์„œ
15:00
it thinks everything might look like a cat.
240
900821
2926
๋ญ๋“ ์ง€ ๊ณ ์–‘์ด๋กœ ๋ณด์ด๋Š”์ง€๋„ ๋ชจ๋ฅด์ฃ .
15:05
(Video) Computer: A young boy is holding a baseball bat.
241
905317
2864
(์ปดํ“จํ„ฐ) "์–ด๋ฆฐ ์†Œ๋…„์ด ์•ผ๊ตฌ ๋ฐฉ๋ง์ด๋ฅผ ๋“ค๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค."
15:08
(Laughter)
242
908181
1765
(์›ƒ์Œ)
15:09
FFL: Or, if it hasn't seen a toothbrush, it confuses it with a baseball bat.
243
909946
4583
์นซ์†”์„ ๋ณธ ์ ์ด ์—†๋‹ค๋ฉด ์•ผ๊ตฌ ๋ฐฉ๋ง์ด์™€ ํ˜ผ๋™ํ•ฉ๋‹ˆ๋‹ค.
15:15
(Video) Computer: A man riding a horse down a street next to a building.
244
915309
3434
(์ปดํ“จํ„ฐ) "๋‚จ์ž๊ฐ€ ๋ง์„ ํƒ€๊ณ  ๊ฑด๋ฌผ ์˜† ๊ธธ์„ ๋‚ด๋ ค๊ฐ‘๋‹ˆ๋‹ค."
15:18
(Laughter)
245
918743
2023
(์›ƒ์Œ)
15:20
FFL: We haven't taught Art 101 to the computers.
246
920766
3552
์šฐ๋ฆฌ๋Š” ์ปดํ“จํ„ฐ์—๊ฒŒ ๋ฏธ์ˆ ์„ ๊ฐ€๋ฅด์น˜์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค.
15:25
(Video) Computer: A zebra standing in a field of grass.
247
925768
2884
(์ปดํ“จํ„ฐ) "์–ผ๋ฃฉ๋ง์ด ์ดˆ์›์— ์„œ์žˆ์Šต๋‹ˆ๋‹ค"
15:28
FFL: And it hasn't learned to appreciate the stunning beauty of nature
248
928652
3367
์ปดํ“จํ„ฐ๋Š” ์ž์—ฐ์˜ ๊ฒฝ์ด๋กœ์šด ์•„๋ฆ„๋‹ค์›€์— ๊ฐ์ƒํ•˜๋Š” ๊ฒƒ์„
15:32
like you and I do.
249
932019
2438
๋ฐฐ์šฐ์ง€๋„ ์•Š์•˜์Šต๋‹ˆ๋‹ค.
15:34
So it has been a long journey.
250
934457
2832
์ด๋Š” ์˜ค๋žœ ์—ฌ์ •์ด์—ˆ์Šต๋‹ˆ๋‹ค.
15:37
To get from age zero to three was hard.
251
937289
4226
0์„ธ์—์„œ 3์„ธ๊นŒ์ง€ ๊ฐ€๋Š” ๊ฑด ํž˜๋“ค์—ˆ์Šต๋‹ˆ๋‹ค.
15:41
The real challenge is to go from three to 13 and far beyond.
252
941515
5596
ํ•˜์ง€๋งŒ ์ง„์งœ ๋„์ „์€ 3์„ธ์—์„œ 13์„ธ, ๊ทธ ์ด์ƒ์œผ๋กœ ๋‚˜์•„๊ฐ€๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
15:47
Let me remind you with this picture of the boy and the cake again.
253
947111
4365
์ด ์†Œ๋…„๊ณผ ์ผ€์ดํฌ์˜ ์‚ฌ์ง„์„ ๋‹ค์‹œ ๋ณด์‹œ์ฃ .
15:51
So far, we have taught the computer to see objects
254
951476
4064
์ง€๊ธˆ๊นŒ์ง€ ์šฐ๋ฆฌ๋Š” ์ปดํ“จํ„ฐ์— ์‚ฌ๋ฌผ์„ ์‹๋ณ„ํ•˜๊ณ 
15:55
or even tell us a simple story when seeing a picture.
255
955540
4458
๊ฐ„๋‹จํ•œ ๋ง์„ ํ•˜๋Š” ๊ฒƒ์„ ๊ฐ€๋ฅด์ณค์Šต๋‹ˆ๋‹ค.
15:59
(Video) Computer: A person sitting at a table with a cake.
256
959998
3576
(์ปดํ“จํ„ฐ) "ํ•œ ์‚ฌ๋žŒ์ด ์ผ€์ดํฌ๊ฐ€ ์žˆ๋Š” ํ…Œ์ด๋ธ”์— ์•‰์•„ ์žˆ์Šต๋‹ˆ๋‹ค."
16:03
FFL: But there's so much more to this picture
257
963574
2630
๊ทธ๋Ÿฌ๋‚˜ ์ด ์‚ฌ์ง„์—๋Š” ์‚ฌ๋žŒ๊ณผ ์ผ€์ดํฌ ์ด์™ธ์—
16:06
than just a person and a cake.
258
966204
2270
๋” ๋งŽ์€ ๊ฒƒ์ด ๋“ค์–ด์žˆ์ฃ .
16:08
What the computer doesn't see is that this is a special Italian cake
259
968474
4467
์ปดํ“จํ„ฐ๊ฐ€ ๋ณด์ง€ ๋ชปํ•˜๋Š” ๊ฒƒ์€ ์ด ํŠน๋ณ„ํ•œ ์ดํƒœ๋ฆฌ ์ผ€์ดํฌ๊ฐ€
16:12
that's only served during Easter time.
260
972941
3217
๋ถ€ํ™œ์ ˆ์—๋งŒ ๋จน๋Š” ๊ฒƒ์ด๋ž€ ๊ฒ๋‹ˆ๋‹ค.
16:16
The boy is wearing his favorite t-shirt
261
976158
3205
์†Œ๋…„์€ ์ž๊ธฐ๊ฐ€ ์ข‹์•„ํ•˜๋Š” ํ‹ฐ์…”์ธ ๋ฅผ ์ž…๊ณ  ์žˆ๋Š”๋ฐ
16:19
given to him as a gift by his father after a trip to Sydney,
262
979363
3970
์•„์ด ์•„๋ฒ„์ง€๊ฐ€ ์‹œ๋“œ๋‹ˆ ์—ฌํ–‰์„ ๋‹ค๋…€์™€ ์„ ๋ฌผ๋กœ ์ค€ ๊ฒƒ์ž…๋‹ˆ๋‹ค.
16:23
and you and I can all tell how happy he is
263
983333
3808
์—ฌ๋Ÿฌ๋ถ„๊ณผ ์ €๋Š” ์ด ์•„์ด๊ฐ€ ์–ผ๋งˆ๋‚˜ ๊ธฐ๋ปํ•˜๋Š”์ง€,
16:27
and what's exactly on his mind at that moment.
264
987141
3203
์ € ์ˆœ๊ฐ„ ๋ฌด์Šจ ์ƒ๊ฐ์„ ํ•˜๋Š”์ง€ ์ด์•ผ๊ธฐํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
16:31
This is my son Leo.
265
991214
3125
์ œ ์•„๋“ค ๋ ˆ์˜ค์ž…๋‹ˆ๋‹ค.
16:34
On my quest for visual intelligence,
266
994339
2624
์‹œ๊ฐ ์ง€๋Šฅ์— ๋Œ€ํ•œ ํƒ๊ตฌ๋ฅผ ํ•˜๋ฉฐ
16:36
I think of Leo constantly
267
996963
2391
์ €๋Š” ํ•ญ์ƒ ๋ ˆ์˜ค์™€
16:39
and the future world he will live in.
268
999354
2903
๋ ˆ์˜ค๊ฐ€ ์‚ด ๋ฏธ๋ž˜์„ธ๊ณ„๋ฅผ ์ƒ๊ฐํ•ฉ๋‹ˆ๋‹ค.
16:42
When machines can see,
269
1002257
2021
๊ธฐ๊ณ„๊ฐ€ ์ธ์‹์„ ํ•˜๊ฒŒ ๋˜๋ฉด,
16:44
doctors and nurses will have extra pairs of tireless eyes
270
1004278
4712
์˜์‚ฌ์™€ ๊ฐ„ํ˜ธ์‚ฌ๋Š” ์‰ฌ์ง€ ์•Š๋Š” ๊ธฐ๊ณ„ ๋ˆˆ์„ ์ด์šฉํ•ด
16:48
to help them to diagnose and take care of patients.
271
1008990
4092
ํ™˜์ž๋ฅผ ์ง„๋‹จํ•˜๊ณ  ๋Œ๋ณผ ์ˆ˜ ์žˆ๊ฒ ์ง€์š”.
16:53
Cars will run smarter and safer on the road.
272
1013082
4383
์ž๋™์ฐจ๋Š” ๋” ๋˜‘๋˜‘ํ•˜๊ณ  ์•ˆ์ „ํ•˜๊ฒŒ ๋„๋กœ๋ฅผ ์ฃผํ–‰ํ•  ๊ฒ๋‹ˆ๋‹ค.
16:57
Robots, not just humans,
273
1017465
2694
์ธ๊ฐ„ ๋ฟ ์•„๋‹ˆ๋ผ ๋กœ๋ด‡์ด
17:00
will help us to brave the disaster zones to save the trapped and wounded.
274
1020159
4849
์žฌ๋‚œ ์ง€์—ญ์—์„œ ๊ฐ‡ํžˆ๊ณ  ๋ถ€์ƒ๋‹นํ•œ ์‚ฌ๋žŒ์„ ๊ตฌํ•˜๋Š” ๊ฑธ ๋„์šธ ๊ฒ๋‹ˆ๋‹ค.
17:05
We will discover new species, better materials,
275
1025798
3796
์šฐ๋ฆฌ๋Š” ๊ธฐ๊ณ„์˜ ๋„์›€์œผ๋กœ ์ƒˆ๋กœ์šด ์ข…, ๋” ๋‚˜์€ ๋ฌผ์งˆ์„ ๋ฐœ๊ฒฌํ•˜๊ณ 
17:09
and explore unseen frontiers with the help of the machines.
276
1029594
4509
๋ณด์ง€ ๋ชปํ•œ ๊ฐœ์ฒ™์ง€๋ฅผ ํƒํ—˜ํ•˜๊ฒŒ ๋  ๊ฒ๋‹ˆ๋‹ค.
17:15
Little by little, we're giving sight to the machines.
277
1035113
4167
์กฐ๊ธˆ์”ฉ ์šฐ๋ฆฌ๋Š” ๊ธฐ๊ณ„์—๊ฒŒ ์‹œ๊ฐ์„ ์ฃผ๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
17:19
First, we teach them to see.
278
1039280
2798
์ฒ˜์Œ์— ์šฐ๋ฆฌ๋Š” ๊ธฐ๊ณ„์—๊ฒŒ ๋ณด๋Š” ๊ฒƒ์„ ๊ฐ€๋ฅด์ณค์Šต๋‹ˆ๋‹ค.
17:22
Then, they help us to see better.
279
1042078
2763
๋‹ค์Œ์—”, ๊ธฐ๊ณ„๊ฐ€ ์šฐ๋ฆฌ๋ฅผ ๋„์™€ ๋” ์ž˜ ๋ณด๊ฒŒ ํ•  ๊ฒ๋‹ˆ๋‹ค.
17:24
For the first time, human eyes won't be the only ones
280
1044841
4165
์ฒ˜์Œ์œผ๋กœ, ์ธ๊ฐ„์˜ ๋ˆˆ์ด ์•„๋‹Œ ๊ฒƒ์ด
17:29
pondering and exploring our world.
281
1049006
2934
์„ธ๊ณ„๋ฅผ ์ƒ๊ฐํ•˜๊ณ  ํƒํ—˜ํ•˜๊ฒŒ ๋˜์—ˆ์Šต๋‹ˆ๋‹ค.
17:31
We will not only use the machines for their intelligence,
282
1051940
3460
์šฐ๋ฆฌ๋Š” ์ธ๊ณต์ง€๋Šฅ ๋•Œ๋ฌธ์— ๊ธฐ๊ณ„๋ฅผ ์ด์šฉํ•  ๋ฟ๋งŒ ์•„๋‹ˆ๋ผ
17:35
we will also collaborate with them in ways that we cannot even imagine.
283
1055400
6179
์ƒ์ƒ์น˜ ๋ชปํ–ˆ๋˜ ๋ฐฉ์‹์œผ๋กœ ๊ธฐ๊ณ„์™€ ํ˜‘๋ ฅํ•˜๊ฒŒ ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
17:41
This is my quest:
284
1061579
2161
์ด๊ฒƒ์ด ์ œ ํƒ๊ตฌ์ž…๋‹ˆ๋‹ค.
17:43
to give computers visual intelligence
285
1063740
2712
์ปดํ“จํ„ฐ์— ์‹œ๊ฐ ์ง€๋Šฅ์„ ๋ถ€์—ฌํ•˜๋Š” ๊ฒƒ,
17:46
and to create a better future for Leo and for the world.
286
1066452
5131
๊ทธ๋ฆฌ๊ณ  ๋ ˆ์˜ค์™€ ์„ธ๊ณ„๋ฅผ ์œ„ํ•ด์„œ ๋” ๋‚˜์€ ๋ฏธ๋ž˜๋ฅผ ๋งŒ๋“œ๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
17:51
Thank you.
287
1071583
1811
๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
17:53
(Applause)
288
1073394
3785
(๋ฐ•์ˆ˜)
์ด ์›น์‚ฌ์ดํŠธ ์ •๋ณด

์ด ์‚ฌ์ดํŠธ๋Š” ์˜์–ด ํ•™์Šต์— ์œ ์šฉํ•œ YouTube ๋™์˜์ƒ์„ ์†Œ๊ฐœํ•ฉ๋‹ˆ๋‹ค. ์ „ ์„ธ๊ณ„ ์ตœ๊ณ ์˜ ์„ ์ƒ๋‹˜๋“ค์ด ๊ฐ€๋ฅด์น˜๋Š” ์˜์–ด ์ˆ˜์—…์„ ๋ณด๊ฒŒ ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค. ๊ฐ ๋™์˜์ƒ ํŽ˜์ด์ง€์— ํ‘œ์‹œ๋˜๋Š” ์˜์–ด ์ž๋ง‰์„ ๋”๋ธ” ํด๋ฆญํ•˜๋ฉด ๊ทธ๊ณณ์—์„œ ๋™์˜์ƒ์ด ์žฌ์ƒ๋ฉ๋‹ˆ๋‹ค. ๋น„๋””์˜ค ์žฌ์ƒ์— ๋งž์ถฐ ์ž๋ง‰์ด ์Šคํฌ๋กค๋ฉ๋‹ˆ๋‹ค. ์˜๊ฒฌ์ด๋‚˜ ์š”์ฒญ์ด ์žˆ๋Š” ๊ฒฝ์šฐ ์ด ๋ฌธ์˜ ์–‘์‹์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ฌธ์˜ํ•˜์‹ญ์‹œ์˜ค.

https://forms.gle/WvT1wiN1qDtmnspy7