英语家园

 找回密码
 注册

QQ登录

只需一步,快速开始

超棒的英语学习网站快速提高英语水平YY语音频道,免费学习本广告位招租

社区广播台

查看: 91|回复: 0
收起左侧

[TED] 【TED】大数据中蕴含着深意的人物画像

[复制链接]

汇报天数: 101 天

连续汇报: 2 天

[LV.6]常住居民II

积分排名 2

管理员

Rank: 45Rank: 45Rank: 45Rank: 45Rank: 45

白雪公主管理员勋章

发表于 2017-10-10 17:42:26 | 显示全部楼层 |阅读模式



00:12

So I'm an artist, but a little bit of a peculiar one. I don't paint. I can't draw. My shop teacher in high school wrote that I was a menace on my report card. You probably don't really want to see my photographs. But there is one thing I know how to do: I know how to program a computer. I can code.

00:34

And people will tell me that 100 years ago, folks like me didn't exist, that it was impossible, that art made with data is a new thing, it's a product of our age, it's something that's really important to think of as something that's very "now." And that's true.

00:52

But there is an art form that's been around for a very long time that's really about using information, abstract information, to make emotionally resonant pieces. And it's called music. We've been making music for tens of thousands of years, right? And if you think about what music is -- notes and chords and keys and harmonies and melodies -- these things are algorithms. These things are systems that are designed to unfold over time, to make us feel. I came to the arts through music. I was trained as a composer, and about 15 years ago, I started making pieces that were designed to look at the intersection between sound and image, to use an image to unveil a musical structure or to use a sound to show you something interesting about something that's usually pictorial.

01:40

So what you're seeing on the screen is literally being drawn by the musical structure of the musicians onstage, and there's no accident that it looks like a plant, because the underlying algorithmic biology of the plant is what informed the musical structure in the first place. So once you know how to do this, once you know how to code with media, you can do some pretty cool stuff.

02:02

This is a project I did for the Sundance Film Festival. Really simple idea: you take every Academy Award Best Picture, you speed it up to one minute each and string them all together. And so in 75 minutes, I can show you the history of Hollywood cinema. And what it really shows you is the history of editing in Hollywood cinema. So on the left, we've got Casablanca; on the right, we've got Chicago. And you can see that Casablanca is a little easier to read. That's because the average length of a cinematic shot in the 1940s was 26 seconds, and now it's around six seconds.

02:39

This is a project that was inspired by some work that was funded by the US Federal Government in the early 2000s, to look at video footage and find a specific actor in any video. And so I repurposed this code to train a system on one person in our culture who would never need to be surveilled in that manner, which is Britney Spears. I downloaded 2,000 paparazzi photos of Britney Spears and trained my computer to find her face and her face alone. I can run any footage of her through it and will center her eyes in the frame, and this sort of is a little double commentary about surveillance in our society. We are very fraught with anxiety about being watched, but then we obsess over celebrity.

03:24

What you're seeing on the screen here is a collaboration I did with an artist named Lián Amaris. What she did is very simple to explain and describe, but very hard to do. She took 72 minutes of activity, getting ready for a night out on the town, and stretched it over three days and performed it on a traffic island in slow motion in New York City. I was there, too, with a film crew. We filmed the whole thing, and then we reversed the process, speeding it up to 72 minutes again, so it looks like she's moving normally and the whole world is flying by.

03:59

At a certain point, I figured out that what I was doing was making portraits. When you think about portraiture, you tend to think about stuff like this. The guy on the left is named Gilbert Stuart. He's sort of the first real portraitist of the United States. And on the right is his portrait of George Washington from 1796. This is the so-called Lansdowne portrait. And if you look at this painting, there's a lot of symbolism, right? We've got a rainbow out the window. We've got a sword. We've got a quill on the desk. All of these things are meant to evoke George Washington as the father of the nation.

04:31

This is my portrait of George Washington. And this is an eye chart, only instead of letters, they're words. And what the words are is the 66 words in George Washington's State of the Union addresses that he uses more than any other president. So "gentlemen" has its own symbolism and its own rhetoric. And it's really kind of significant that that's the word he used the most. This is the eye chart for George W. Bush, who was president when I made this piece. And how you get there, from "gentlemen" to "terror" in 43 easy steps, tells us a lot about American history, and gives you a different insight than you would have looking at a series of paintings. These pieces provide a history lesson of the United States through the political rhetoric of its leaders. Ronald Reagan spent a lot of time talking about deficits. Bill Clinton spent a lot of time talking about the century in which he would no longer be president, but maybe his wife would be. Lyndon Johnson was the first President to give his State of the Union addresses on prime-time television; he began every paragraph with the word "tonight." And Richard Nixon, or more accurately, his speechwriter, a guy named William Safire, spent a lot of time thinking about language and making sure that his boss portrayed a rhetoric of honesty.

05:51

This project is shown as a series of monolithic sculptures. It's an outdoor series of light boxes. And it's important to note that they're to scale, so if you stand 20 feet back and you can read between those two black lines, you have 20/20 vision.

06:05

This is a portrait. And there's a lot of these. There's a lot of ways to do this with data. I started looking for a way to think about how I can do a more democratic form of portraiture, something that's more about my country and how it works. Every 10 years, we make a census in the United States. We literally count people, find out who lives where, what kind of jobs we've got, the language we speak at home. And this is important stuff -- really important stuff. But it doesn't really tell us who we are. It doesn't tell us about our dreams and our aspirations.

06:39

And so in 2010, I decided to make my own census. And I started looking for a corpus of data that had a lot of descriptions written by ordinary Americans. And it turns out that there is such a corpus of data that's just sitting there for the taking. It's called online dating.

06:56

So in 2010, I joined 21 different online dating services, as a gay man, a straight man, a gay woman and a straight woman, in every zip code in America and downloaded about 19 million people's dating profiles -- about 20 percent of the adult population of the United States. I have obsessive-compulsive disorder. This is going to become really freaking obvious. Just go with me.

07:19

So what I did was I sorted all this stuff by zip code. And I looked at word analysis. These are some dating profiles from 2010 with the word "lonely" highlighted. If you look at these things topographically, if you imagine dark colors to light colors are more use of the word, you can see that Appalachia is a pretty lonely place. You can also see that Nebraska ain't that funny. This is the kinky map, so what this is showing you is that the women in Alaska need to get together with the men in southern New Mexico, and have a good time. And I have this at a pretty granular level, so I can tell you that the men in the eastern half of Long Island are way more interested in being spanked than men in the western half of Long Island. This will be your one takeaway from this whole conference. You're going to remember that fact for, like, 30 years.

08:20

When you bring this down to a cartographic level, you can make maps and do the same trick I was doing with the eye charts. You can replace the name of every city in the United States with the word people use more in that city than anywhere else. If you've ever dated anyone from Seattle, this makes perfect sense. You've got "pretty." You've got "heartbreak." You've got "gig." You've got "cigarette." They play in a band and they smoke. And right above that you can see "email." That's Redmond, Washington, which is the headquarters of the Microsoft Corporation. Some of these you can guess -- so, Los Angeles is "acting" and San Francisco is "gay." Some are a little bit more heartbreaking. In Baton Rouge, they talk about being curvy; downstream in New Orleans, they still talk about the flood. Folks in the American capital will say they're interesting. People in Baltimore, Maryland, will say they're afraid. This is New Jersey. I grew up somewhere between "annoying" and "cynical."

09:15

And New York City's number one word is "now," as in, "Now I'm working as a waiter, but actually I'm an actor."

09:22

Or, "Now I'm a professor of engineering at NYU, but actually I'm an artist." If you go upstate, you see "dinosaur." That's Syracuse. The best place to eat in Syracuse, New York, is a Hell's Angels barbecue joint called Dinosaur Barbecue. That's where you would take somebody on a date. I live somewhere between "unconditional" and "midsummer," in Midtown Manhattan. And this is gentrified North Brooklyn, so you've got "DJ" and "glamorous" and "hipsters" and "urbane." So that's maybe a more democratic portrait. And the idea was, what if we made red-state and blue-state maps based on what we want to do on a Friday night?

09:53

This is a self-portrait. This is based on my email, about 500,000 emails sent over 20 years. You can think of this as a quantified selfie. So what I'm doing is running a physics equation based on my personal data. You have to imagine everybody I've ever corresponded with. It started out in the middle and it exploded with a big bang. And everybody has gravity to one another, gravity based on how much they've been emailing, who they've been emailing with. And it also does sentimental analysis, so if I say "I love you," you're heavier to me. And you attract to my email addresses in the middle, which act like mainline stars. And all the names are handwritten.

10:30

Sometimes you do this data and this work with real-time data to illuminate a specific problem in a specific city. This is a Walther PPK 9mm semiautomatic handgun that was used in a shooting in the French Quarter of New Orleans about two years ago on Valentine's Day in an argument over parking. Those are my cigarettes. This is the house where the shooting took place. This project involved a little bit of engineering. I've got a bike chain rigged up as a cam shaft, with a computer driving it. That computer and the mechanism are buried in a box. The gun's on top welded to a steel plate. There's a wire going through to the trigger, and the computer in the box is online. It's listening to the 911 feed of the New Orleans Police Department, so that anytime there's a shooting reported in New Orleans,

11:12

(Gunshot sound)

11:13

the gun fires. Now, there's a blank, so there's no bullet. There's big light, big noise and most importantly, there's a casing. There's about five shootings a day in New Orleans, so over the four months this piece was installed, the case filled up with bullets. You guys know what this is -- you call this "data visualization." When you do it right, it's illuminating. When you do it wrong, it's anesthetizing. It reduces people to numbers. So watch out.

11:44

One last piece for you. I spent the last summer as the artist in residence for Times Square. And Times Square in New York is literally the crossroads of the world. One of the things people don't notice about it is it's the most Instagrammed place on Earth. About every five seconds, someone commits a selfie in Times Square. That's 17,000 a day, and I have them all.

12:08

These are some of them with their eyes centered.

12:11

Every civilization, will use the maximum level of technology available to make art. And it's the responsibility of the artist to ask questions about what that technology means and how it reflects our culture.

12:21

So I leave you with this: we're more than numbers. We're people, and we have dreams and ideas. And reducing us to statistics is something that's done at our peril.

12:30

Thank you very much.00:12

我是一个艺术家 但有点不同 我不会油画 我也不会速写 我的高中劳技老师曾在我的评语里写 我是一个捣蛋鬼 你可能不会想看我的摄影 但有一件事 我知道怎么做好 我会给计算机设计程序 我会编码

00:34

人们会告诉我 在一百年前 不存在像我这样的人 我做的事情是不可能实现的 这种用字符创作的艺术是全新的 是我们新时代的产物 作为一种最新的产物的代表 它也格外重要 这都没错

00:52

但有一种存在了很久的艺术形式 也是利用信息 利用抽象的信息 产生情感上的共鸣 它就是音乐 我们创作音乐已有上千年了 对吧 如果要说音乐是什么 音乐就是音符 和弦 音调 和声和旋律 这些东西就是算法 它们是设计好的系统 会随着时间的推移打开 让我们能去感受 我是通过音乐接触艺术的 我曾被训练为作曲家 十五年前 我开始创作一些片段 希望能达到 声音与图像的完美融合 用一幅画面去揭示一个音乐结构 或者用一个声音去展示有趣的事情 那些一般比较形象化的东西

01:40

所以你在屏幕上看到的实际上都是 由舞台上的音乐家的乐曲结构构成的

01:47

它们不出意外的看起来像是植物 因为它最初的乐曲结构本质上就是 植物最基本的生物算法 所以当你知道如何去编程时 你就可以做出很多有意思的东西

02:02

这是我曾为圣丹斯电影节做过的一个项目 很简单的想法就是 收集所有奥斯卡得奖摄影 然后把每一部加速播放到一分钟 再把他们都串在一起 所以我可以只用75分钟 来向你展示好莱坞电影史 而真正展示的是 好莱坞电影的编辑史 所以我们可以得到 左边的是卡萨布兰卡市 右边的是芝加哥 你也可以发现卡萨布兰卡市看起来更容易 那是因为在20世纪40年代 电影镜头的平均长度为26秒 而如今约为6秒

02:39

这个项目是受启发于 美国联邦政府在21世纪初 赞助的一些工作 就是从任何片段录像镜头中 找出一个具体的人物 我改写了程序来 训练这个系统 去识别一个在我们的文化中 永远不需要被这样监视的人物 她就是布兰妮•斯皮尔斯(Britney Spears) 我下载了2000张狗仔队偷拍的布兰妮的照片 训练我的电脑去识别她 而且仅仅是她 我可以播放任意她的片段 电脑会在相框内聚焦她的眼睛

03:13

这个事情反映了 我们社会对于监控的一种矛盾心理 我们因为被监视着而充满了焦虑 却时刻对公众人物做着这样的事情

03:24

你现在在屏幕上看到的 是我和一名叫Lián Amaris的艺术家共同完成的 她的任务说起来容易 做起来难 她需要呈现的在是72分钟里 为一次城里的狂欢夜梳妆打扮 但要拉长到整整三天 她需要在纽约街头的一个交通岛 用慢动作来展现这一切 我也在那里 带着一个摄影团队 我们录下了整个过程 然后把它倒放 再加速到72分钟 使她看起来是以正常速度移动 整个世界却是在飞速的运转着

03:59

有一次 我突然发觉 我正在做的正是塑造画像 当提到肖像画时 你可能会想到这样的东西 左边的这个人叫吉伯特•斯图尔特 他可能是美国第一位肖像画家 右边的就是他在1796年为乔治•华盛顿做的肖像 这就是所谓的“兰丝唐呢绒”肖像 如果你仔细看这幅画 会发现很多象征符号 对吗 这里有窗外的彩虹 这里有剑 这里有桌上的羽毛笔 所有这些事物都在暗示我们 这是乔治•华盛顿——美国之父

04:31

这个是我对乔治•华盛顿的肖像画 它像是一个视力测试表 只是它不是由字母组成 而是单词 这66个单词是 乔治•华盛顿在国情咨文演说中 比其他总统用的都要多的单词 所以“先生们”这个词 有其独特的象征和修辞意义 作为他用得最频繁的词 这个词是有着某种重要意义的 这是乔治•布什的“视力表” 当我做这张表的时候 他还不是总统 从“先生们”到“恐怖行动” 经历了43个简单的步骤 (43届总统) 蕴藏着很多关于美国的历史 相比于看一系列的画作 可以使你获得别样的体验 这些作品通过领导人的政治辞令 为我们展现了一节生动的美国历史课 罗纳德•里根花了很多时间谈论财政赤字 比尔•克林顿花了很多时间谈论 他不再是总统的21世纪 她的夫人有可能成为总统的21世纪 林登•约翰逊是第一任 在电视黄金时间发表国情咨文演说的总统 他每段都以“今晚”开头 理查德•尼克松 更准确的说 他的撰稿人 一个叫威廉•萨菲尔的人 花了很多时间斟酌他的言辞 以确保他的老板正在演说着“诚实”的辞令

05:51

这个项目以一系列统一的雕塑来展现 是一系列的户外灯箱 值得注意的是 它们和视力表一样是成比例的 也就是说 如果你站在20尺外 能领会到这两行黑字的言外之意 说明你视力正常

06:05

这是一种肖像画 类似的还有很多 有很多种方法来利用这些数据 我开始寻找一种方法 用一种更大众的形式来表现 更多的关于我的国家和它的运转方式 美国每十年人口普查一次 我们做的就是清点人数 查清谁住在哪 我们都有什么工作 我们在家都说什么语言 的确 这些都是非常重要的事 但并不能描述出我们的真实身份 这些不能描述我们的梦想 我们的和抱负

06:39

因此 在2010年 我决定做我自己的人口普查 我开始寻找那些 由普通美国群众写的有大量描述性内容的数据库 结果我发现 就有着这样的数据库 等着我去利用起来 就是在线约会网站

06:56

所以 2010年我注册了21个不同的在线约会网站 以各种不同的身份 包括同性恋男 直男 同性恋女 直女 还用美国各地的邮编 下载了一千九百万人的档案 也就是大约20%美国成人人口 我有强迫症 会越来越明显的 稍微忍忍吧

07:19

所以我把所有资料按邮编分类 用词汇分析法来分析 这是一些2010年的档案中 包含“孤独”这个词的被我高亮的出来 如果从地理的角度来看 颜色越亮代表词语使用量越高 你可以发现阿巴拉契亚是一个十分孤独的地方 还可以看出内布拉斯加州也不是那么的有趣 这是一幅有些扭曲的地图 它告诉我们的是 阿拉斯加的女人需要和 南方新墨西哥州的男人在一起 过上幸福生活 我还可以告诉你一些更细微的发现 从长岛东半边来的男人 比西半边的 更乐于被挨巴掌 这可以作为你在这次活动的收获之一 这个结论你应该可以记大概 30年吧

08:20

把这个想法应用到地图上 同样可以实现 就像我做的“视力表”一样 你可以把美国每个城市的名字 替换成每个地方的人们使用频率高的词语 如果你曾和来自西雅图的人约过会 你就会明白 你会听到“好漂亮” 你会听到“心碎了” 你会听到“现场演出“ 你会听到“香烟” 他们会有乐队的演出 他们也抽烟 右上方你可以看到“电子邮件” 那是华盛顿州的雷蒙德市 正是微软总部的所在地 有些你都可以猜到 比如 洛杉矶是“表演” 旧金山是“同性恋” 有些是有点令人悲伤的 在巴吞鲁日 人们谈论肥胖 到下游的新奥尔良市 人们仍在谈论洪水 在美国首都的人们会说他们很有趣 在马里兰州巴尔的摩的人们则会说他们很害怕 这是新泽西 我在“烦人“和”愤世嫉俗“之间长大

09:15

纽约市排名第一的词是“现在” 比如 别看我现在是个服务生 实际上我是个演员

09:22

或者 “我现在是个纽约大学的工程教授 但实际上我是个艺术家” 往北一些 你会看到“恐龙” 那是锡拉丘兹 纽约州的锡拉丘兹里最好的用餐地点 是一个叫恐龙烧烤的飞车党烧烤酒吧 那是一个可以约会的好地方 我住在一个在“无条件的”和“夏至“之间的地方 在曼哈顿中部 这是修缮后的北布鲁克林 所以你会看到“DJ” “迷人的” “嬉皮士”和“都市化的” 所以这应该是个更大众的肖像 另一个想法就是 把这个应用到红蓝两党的地图上 基于周五晚上我们想做什么

09:53

这将是一副自画像 基于我的电子邮箱 有这20年里的大约五十万封邮件 你可以把这个想象成一个定量的自拍照 我做的就是在我的个人数据的基础上 套用一个物理公式 想象一下与我有过交集的每一个人 从中心开始 在一声巨响后向四面八方炸裂 而每个人之间互相都有引力 引力大小取决于他们之间发送电子邮件的频率 以及他们给谁发过电子邮件 这之中也包含了感性分析 例如我说“我爱你” 那么你对我的引力就越大 你就会被我的电子邮件所吸引 也就是在画面中央的 就像是一线明星一样 所有这些名字都是手写的

10:30

有时候你可以这样来处理实时数据 去阐释一个城市里发生的某个具体的问题 这是一把沃尔特-PPK式9毫米自动手枪 它被用于2年前情人节在新奥尔良法国区 一起因停车问题的争吵而引发的枪击事件中 这些是我的香烟 这是那幢发生枪击的房子 这个项目包含一点工程学 我用自行车的链条组装成了一个凸轮轴 由计算机驱动 整个计算机和机械被放在一个盒子里 这把手枪被焊在一个钢铁盘上面 有一根电话线穿过扳机 盒子里的计算机是联网的 它连接着911新奥尔良分局的输入口 因此在新奥尔良只要有枪击案上报

11:13

这把手枪就会开火 现在它是空的 里面没有子弹 会有明亮的灯光和嘈杂的声响 更重要的是 那里有个盒子 在新奥尔良 每天会有五个枪击案 所以在这个装置装好后的四个月 这个盒子就装满了子弹 大家都知道这是什么 你们把它叫做“数据可视化” 你正确的利用它就会很有启发 你若做错了就会让人麻痹大意 它会让生命变成数字 所以务必要当心

11:44

最后一件要跟你们分享的事 去年夏天 我以一名艺术家的身份 驻扎在时代广场 纽约的时代广场算得上是世界的交叉路口 而人们没有注意到的是 它也是全球发Instagram最多的地方 在时代广场 约每五秒钟就有人在一张自拍照下面评论 也就是明天17000条 我全都收集了起来

12:08

这些是人们目视正中心的照片

12:11

每一种文明 都最大程度地利用科技去创造艺术 这也是艺术家的责任去发出疑问 科技意味着什么 科技如何反映我们的文化

12:21

所以我想告诉你们的是 我们不仅仅是数字 我们是人类 我们有梦想和创意 把人简化成数据 是将我们自己至于险境 谢谢 (掌声)

[发帖际遇]: 一个袋子砸在了 katy 头上,katy 赚了 4 元 家元. 幸运榜 / 衰神榜
游客
请先登录
您需要登录后才可以回帖 登录 | 注册

本版积分规则

提示
随便
看看

精彩
图片

帖子
导读
英语家园公众微信
微信扫一扫
关注英语家园公众微信
快速回复 返回顶部 返回列表