1
00:00:01,317 --> 00:00:03,809
I am an astrophysicist.

2
00:00:03,833 --> 00:00:06,888
I research stellar explosions
across the universe.

3
00:00:07,603 --> 00:00:08,803
But I have a flaw:

4
00:00:09,325 --> 00:00:11,808
I'm restless, and I get bored easily.

5
00:00:12,356 --> 00:00:15,552
And although as an astrophysicist,
I have the incredible opportunity

6
00:00:15,576 --> 00:00:17,508
to study the entire universe,

7
00:00:17,532 --> 00:00:20,880
the thought of doing
only that, always that,

8
00:00:20,904 --> 00:00:23,135
makes me feel caged and limited.

9
00:00:24,762 --> 00:00:28,676
What if my issues with
keeping attention and getting bored

10
00:00:28,700 --> 00:00:30,182
were not a flaw, though?

11
00:00:30,206 --> 00:00:32,873
What if I could turn them into an asset?

12
00:00:33,830 --> 00:00:36,475
An astrophysicist cannot
touch or interact with

13
00:00:36,499 --> 00:00:38,037
the things that she studies.

14
00:00:38,061 --> 00:00:41,774
No way to explode a star in a lab
to figure out why or how it blew up.

15
00:00:42,164 --> 00:00:44,694
Just pictures and movies of the sky.

16
00:00:45,339 --> 00:00:47,560
Everything we know about the universe,

17
00:00:47,584 --> 00:00:50,675
from the big bang
that originated space and time,

18
00:00:50,699 --> 00:00:53,493
to the formation and evolution
of stars and galaxies,

19
00:00:53,517 --> 00:00:55,977
to the structure of our own solar system,

20
00:00:56,001 --> 00:00:58,801
we figured out studying images of the sky.

21
00:01:00,006 --> 00:01:03,958
And to study a system
as complex as the entire universe,

22
00:01:03,982 --> 00:01:08,687
astrophysicists are experts
at extracting simple models and solutions

23
00:01:08,711 --> 00:01:11,050
from large and complex data sets.

24
00:01:11,705 --> 00:01:14,130
So what else can I do with this expertise?

25
00:01:16,030 --> 00:01:20,125
What if we turned the camera
around towards us?

26
00:01:21,057 --> 00:01:24,048
At the Urban Observatory,
that is exactly what we are doing.

27
00:01:24,072 --> 00:01:26,564
Greg Dobler, also an astrophysicist

28
00:01:26,588 --> 00:01:27,755
and my husband,

29
00:01:27,779 --> 00:01:31,739
created the first urban observatory
in New York University in 2013,

30
00:01:31,763 --> 00:01:33,319
and I joined in 2015.

31
00:01:33,752 --> 00:01:35,681
Here are some of the things that we do.

32
00:01:36,196 --> 00:01:38,466
We take pictures of the city at night

33
00:01:38,490 --> 00:01:41,079
and study city lights like stars.

34
00:01:41,514 --> 00:01:43,526
By studying how light changes over time

35
00:01:43,550 --> 00:01:45,624
and the color of astronomical lights,

36
00:01:45,648 --> 00:01:48,461
I gain insight about the nature
of exploding stars.

37
00:01:48,934 --> 00:01:51,204
By studying city lights the same way,

38
00:01:51,228 --> 00:01:55,910
we can measure and predict how much energy
the city needs and consumes

39
00:01:55,934 --> 00:01:57,781
and help build a resilient grid

40
00:01:57,805 --> 00:02:01,104
that will support the needs
or growing urban environments.

41
00:02:02,283 --> 00:02:05,640
In daytime images,
we capture plumes of pollution.

42
00:02:06,274 --> 00:02:09,744
Seventy-five percent
of greenhouse gases in New York City

43
00:02:09,768 --> 00:02:13,400
come from a building like this one,
burning oil for heat.

44
00:02:14,477 --> 00:02:16,872
You can measure pollution
with air quality sensors.

45
00:02:16,896 --> 00:02:20,738
But imagine putting a sensor
on each New York City building,

46
00:02:20,762 --> 00:02:23,468
reading in data from a million monitors.

47
00:02:23,492 --> 00:02:24,818
Imagine the cost.

48
00:02:26,048 --> 00:02:29,476
With a team of NYU students,
we built a mathematical model,

49
00:02:29,500 --> 00:02:32,889
a neural network that can detect
and track these plumes

50
00:02:32,913 --> 00:02:34,603
over the New York City skyline.

51
00:02:34,627 --> 00:02:36,109
We can classify them --

52
00:02:36,133 --> 00:02:39,148
harmless steam plumes,
white and evanescent;

53
00:02:39,172 --> 00:02:42,688
polluting smokestacks,
dark and persistent --

54
00:02:42,712 --> 00:02:46,471
and provide policy makers
with a map of neighborhood pollution.

55
00:02:47,777 --> 00:02:51,666
This cross-disciplinary project
created transformational solutions.

56
00:02:53,942 --> 00:02:56,855
But the data analysis methodologies
we use in astrophysics

57
00:02:56,879 --> 00:02:58,840
can be applied to all sorts of data,

58
00:02:58,864 --> 00:03:00,014
not just images.

59
00:03:00,450 --> 00:03:02,934
We were asked to help
a California district attorney

60
00:03:02,958 --> 00:03:06,379
understand prosecutorial delays
in their jurisdiction.

61
00:03:06,839 --> 00:03:09,506
There are people on probation
or sitting in jail,

62
00:03:09,530 --> 00:03:12,141
awaiting for trial sometimes for years.

63
00:03:12,165 --> 00:03:14,609
They wanted to know
what kind of cases dragged on,

64
00:03:14,633 --> 00:03:17,757
and they had a massive data set
to explore to understand it,

65
00:03:17,781 --> 00:03:19,193
but didn't have the expertise

66
00:03:19,217 --> 00:03:21,765
or the instruments
in their office to do so.

67
00:03:21,789 --> 00:03:23,535
And that's where we came in.

68
00:03:23,559 --> 00:03:26,956
I worked with my colleague,
public policy professor Angela Hawken,

69
00:03:26,980 --> 00:03:30,466
and our team first created
a visual dashboard

70
00:03:30,490 --> 00:03:34,390
for DAs to see and better understand
the prosecution process.

71
00:03:34,997 --> 00:03:37,926
But also, we ourselves
analyzed their data,

72
00:03:37,950 --> 00:03:40,470
looking to see if the duration
of the process

73
00:03:40,494 --> 00:03:43,684
suffered from social inequalities
in their jurisdiction.

74
00:03:44,367 --> 00:03:45,909
We did so using methods

75
00:03:45,933 --> 00:03:48,906
that I would use to classify
thousands of stellar explosions,

76
00:03:48,930 --> 00:03:51,585
applied to thousands of court cases.

77
00:03:51,609 --> 00:03:52,760
And in doing so,

78
00:03:52,784 --> 00:03:55,617
we built a model that can be applied
to other jurisdictions

79
00:03:55,641 --> 00:03:57,831
who are willing to explore their biases.

80
00:03:57,855 --> 00:04:01,101
These collaborations between
domain experts and astrophysicists

81
00:04:01,125 --> 00:04:03,180
created transformational solutions

82
00:04:03,204 --> 00:04:05,604
to help improve people's quality of life.

83
00:04:07,426 --> 00:04:08,910
But it is a two-way road.

84
00:04:08,934 --> 00:04:11,477
I bring my astrophysics background
to urban science,

85
00:04:11,501 --> 00:04:15,342
and I bring what I learn in urban science
back to astrophysics.

86
00:04:15,930 --> 00:04:17,782
Light echoes:

87
00:04:18,461 --> 00:04:23,028
the reflections of stellar explosions
onto interstellar dust.

88
00:04:24,046 --> 00:04:29,831
In our images, these reflections appear
as white, evanescent, moving features,

89
00:04:29,855 --> 00:04:31,005
just like plumes.

90
00:04:31,363 --> 00:04:35,258
I am adapting the same models
that detect plumes in city images

91
00:04:35,282 --> 00:04:38,210
to detect light echoes
in images of the sky.

92
00:04:40,290 --> 00:04:43,583
By exploring the things
that interest and excite me,

93
00:04:43,607 --> 00:04:45,403
reaching outside of my domain,

94
00:04:45,427 --> 00:04:48,303
I did turn my restlessness into an asset.

95
00:04:49,031 --> 00:04:54,078
We, you, all have a unique perspective
that can generate new insight

96
00:04:54,102 --> 00:04:58,285
and lead to new, unexpected,
transformational solutions.

97
00:04:58,944 --> 00:05:00,106
Thank you.

98
00:05:00,130 --> 00:05:04,288
(Applause)