{"id":3244,"date":"2019-11-25T11:00:47","date_gmt":"2019-11-25T17:00:47","guid":{"rendered":"https:\/\/alliwalk.com\/blog\/?p=3244"},"modified":"2019-11-24T19:06:19","modified_gmt":"2019-11-25T01:06:19","slug":"this-blog-post-does-not-exist","status":"publish","type":"post","link":"https:\/\/alliwalk.com\/blog\/2019\/11\/this-blog-post-does-not-exist\/","title":{"rendered":"This Blog Post Does Not Exist&#8230;Or Does It?"},"content":{"rendered":"<p>Shout out to those hard-working <a href=\"https:\/\/en.wikipedia.org\/wiki\/Generative_adversarial_network\">generative adversarial network<\/a>\u00a0(GAN)! Ok, I don&#8217;t really know about hard-working, but it is pretty cool. Let&#8217;s review.<\/p>\n<h2>Intro to GAN<\/h2>\n<p>GAN\u00a0technology came out\u00a0in\u00a0internally at Nvidia, the computer graphics company, in <a href=\"https:\/\/arxiv.org\/abs\/1406.2661\" target=\"_blank\" rel=\"noopener noreferrer\">2014<\/a> and released publicly in <a href=\"https:\/\/arxiv.org\/abs\/1812.04948\" target=\"_blank\" rel=\"noopener noreferrer\">2018<\/a>\u00a0(as StyleGAN). The studies are linked.<\/p>\n<p>I&#8217;m not going to pretend I understand the details of either\u00a0study. But based on the Wikipedia article linked above, what I can explain is that a GAN is the result of two\u00a0neural networks\u00a0that compete against\u00a0each\u00a0other, in one of those mathematical, strategy &#8220;games&#8221; researchers like to play.\u00a0 The networks study photographs (or text) and then decide which elements to use to\u00a0generate something new.<\/p>\n<p>I think some people will find\u00a0it creepy, but I think it&#8217;s cool. For the human photos, it&#8217;s hard to not project a humanity into the faces, even though, logically, these are people that have never existed. The generated fake people seem, somehow&#8230;.special.<\/p>\n<p>Here&#8217;s a quick intro to some GAN and some links to explore. Enjoy!<\/p>\n<hr \/>\n<h3>Explore Some Fake Stuff<\/h3>\n<p>As I said above, one of the possible outputs of a GAN can be a surprisingly realistic looking photographs. View some of them at <a href=\"https:\/\/thispersondoesnotexist.com\/\" target=\"_blank\" rel=\"noopener noreferrer\">thispersondoesnotexist.com<\/a>. Every time you refresh, you get a new person that has never existed. It&#8217;s like dreaming for computers.<\/p>\n<p>This <a href=\"https:\/\/github.com\/paubric\/thisrepositorydoesnotexist\" target=\"_blank\" rel=\"noopener noreferrer\">github repository<\/a> has a list of a few more websites\u00a0of GANs that\u00a0generate photographs of humans, fake\u00a0rental ads, articles, anime, and more results\u00a0that don&#8217;t exist. There are fake cats, but computers seem to have trouble with animals. I suggest not looking at them. (They&#8217;re creepy.)<\/p>\n<h3>Test Yourself, Human!<\/h3>\n<div id='gallery-1' class='gallery galleryid-3244 gallery-columns-2 gallery-size-large'><figure class='gallery-item'>\n\t\t\t<div class='gallery-icon landscape'>\n\t\t\t\t<a href='https:\/\/alliwalk.com\/blog\/2019\/11\/this-blog-post-does-not-exist\/woman_1\/'><img width=\"840\" height=\"840\" src=\"https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/Woman_1.jpg\" class=\"attachment-large size-large\" alt=\"\" loading=\"lazy\" aria-describedby=\"gallery-1-3395\" srcset=\"https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/Woman_1.jpg 1024w, https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/Woman_1-150x150.jpg 150w, https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/Woman_1-300x300.jpg 300w, https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/Woman_1-768x768.jpg 768w\" sizes=\"(max-width: 709px) 85vw, (max-width: 909px) 67vw, (max-width: 1362px) 62vw, 840px\" \/><\/a>\n\t\t\t<\/div>\n\t\t\t\t<figcaption class='wp-caption-text gallery-caption' id='gallery-1-3395'>\n\t\t\t\tThis woman does not exist\n\t\t\t\t<\/figcaption><\/figure><figure class='gallery-item'>\n\t\t\t<div class='gallery-icon landscape'>\n\t\t\t\t<a href='https:\/\/alliwalk.com\/blog\/2019\/11\/this-blog-post-does-not-exist\/fireshot-capture-225-new-affordable-family-friendly-house-with-garden-in-brunswick-gu_-thisrentaldoesnotexist-com\/'><img width=\"840\" height=\"591\" src=\"https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/FireShot-Capture-225-New-Affordable-Family-Friendly-HOUSE-with-garden-in-Brunswick-Gu_-thisrentaldoesnotexist.com_-1024x720.png\" class=\"attachment-large size-large\" alt=\"\" loading=\"lazy\" aria-describedby=\"gallery-1-3412\" srcset=\"https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/FireShot-Capture-225-New-Affordable-Family-Friendly-HOUSE-with-garden-in-Brunswick-Gu_-thisrentaldoesnotexist.com_-1024x720.png 1024w, https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/FireShot-Capture-225-New-Affordable-Family-Friendly-HOUSE-with-garden-in-Brunswick-Gu_-thisrentaldoesnotexist.com_-300x211.png 300w, https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/FireShot-Capture-225-New-Affordable-Family-Friendly-HOUSE-with-garden-in-Brunswick-Gu_-thisrentaldoesnotexist.com_-768x540.png 768w, https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/FireShot-Capture-225-New-Affordable-Family-Friendly-HOUSE-with-garden-in-Brunswick-Gu_-thisrentaldoesnotexist.com_-1200x843.png 1200w, https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/FireShot-Capture-225-New-Affordable-Family-Friendly-HOUSE-with-garden-in-Brunswick-Gu_-thisrentaldoesnotexist.com_.png 1440w\" sizes=\"(max-width: 709px) 85vw, (max-width: 909px) 67vw, (max-width: 1362px) 62vw, 840px\" \/><\/a>\n\t\t\t<\/div>\n\t\t\t\t<figcaption class='wp-caption-text gallery-caption' id='gallery-1-3412'>\n\t\t\t\tThe rental ad does not exist.\n\t\t\t\t<\/figcaption><\/figure>\n\t\t<\/div>\n\n<p>Some of these faces look pretty realistic. For instance, the image\u00a0above-left is\u00a0an &#8220;image of a young woman generated by StyleGAN, an generative adversarial network (GAN). The person in this photo does not exist, but is generated by an artificial intelligence based on an analysis of portraits.&#8221; The ad on the right is fake, too. <em>Could you tell?<\/em><\/p>\n<p>If you want to really test yourself, the website <a href=\"http:\/\/www.whichfaceisreal.com\/learn.html\" target=\"_blank\" rel=\"noopener noreferrer\">Which Face Is Real<\/a>\u00a0throws up 2 side-by-side images.<\/p>\n<p><a href=\"https:\/\/flic.kr\/p\/8WpLNn\"><img loading=\"lazy\" src=\"https:\/\/live.staticflickr.com\/5207\/5210226441_a3b15aafbd_c.jpg\" alt=\"Catching up with Friends.\" width=\"800\" height=\"533\" \/><\/a><\/p>\n<p>Unless you&#8217;re a <del>dog<\/del>\u00a0cat on the internet, you can probably tell the difference (and <del>probably<\/del>\u00a0<del>even<\/del>\u00a0<em>especially<\/em> if you are a cat). The fake images have a few tell-tale signs like smudged backgrounds, odd looking teeth,\u00a0unusual wrinkles, and some of them just don&#8217;t seem right.<\/p>\n<p>You can also test yourself with the fake poems at <a href=\"http:\/\/botpoet.com\/vote\/sign-post\/\">Bot Poet<\/a>.<\/p>\n<h3>Offshoots<\/h3>\n<p><strong>Composites<\/strong> &#8211; There are some offshoots of the original technology, like this\u00a0<a href=\"https:\/\/github.com\/pbaylies\/stylegan-encoder\" target=\"_blank\" rel=\"noopener noreferrer\">github repository<\/a>\u00a0which shows\u00a0an example of a composite\u00a0generated image using a StyleGAN image + an image of the Mona Lisa.<\/p>\n<p><strong>Digital &#8220;models&#8221;<\/strong> &#8211; I wouldn&#8217;t say these examples are quite the same, but there are already digital &#8220;models&#8221; on Instagram and in advertising campaigns.\u00a0<a href=\"https:\/\/www.instagram.com\/lilmiquela\/\" target=\"_blank\" rel=\"noopener noreferrer\">Lil Miquela<\/a> has over 1.5 million followers on Instagram, and there&#8217;s a <a href=\"https:\/\/www.thediigitals.com\/\" target=\"_blank\" rel=\"noopener noreferrer\">modeling agency<\/a> specializing in digital &#8220;models&#8221;.<\/p>\n<p><strong>Video<\/strong> &#8211; The one below is uses around 8 frames of video to train their neural network, resulting in a real-time talking head model. It could be like one of those &#8220;deep-fake&#8221; videos, except the new heads are people that don&#8217;t exist. And you can kind of tell, if you watch the video below,\u00a0human, the new heads do not really look too realistic (yet).<\/p>\n<p><em><strong>Few-Shot Adversarial Learning of Realistic Neural Talking Head Models<\/strong><\/em><\/p>\n<p><iframe loading=\"lazy\" width=\"840\" height=\"473\" src=\"https:\/\/www.youtube.com\/embed\/p1b5aiTrGzY?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen><\/iframe><\/p>\n<p style=\"padding-left: 30px;\">Statement regarding the purpose and effect of the technology<br \/>\n(NB: this statement reflects personal opinions of the authors and not of their organizations)<\/p>\n<p style=\"padding-left: 30px;\">We believe that telepresence technologies in AR, VR and other media are to transform the world in the not-so-distant future. Shifting a part of human life-like communication to the virtual and augmented worlds will have several positive effects. It will lead to a reduction in long-distance travel and short-distance commute. It will democratize education, and improve the quality of life for people with disabilities. It will distribute jobs more fairly and uniformly around the World. It will better connect relatives and friends separated by distance. To achieve all these effects, we need to make human communication in AR and VR as realistic and compelling as possible, and the creation of photorealistic avatars is one (small) step towards this future. In other words, in future telepresence systems, people will need to be represented by the realistic semblances of themselves, and creating such avatars should be easy for the users. This application and scientific curiosity is what drives the research in our group, including the project presented in this video.<\/p>\n<p style=\"padding-left: 30px;\">We realize that our technology can have a negative use for the so-called \u201cdeepfake\u201d videos. However, it is important to realize, that Hollywood has been making fake videos (aka \u201cspecial effects\u201d) for a century, and deep networks with similar capabilities have been available for the past several years (see links in the paper). Our work (and quite a few parallel works) will lead to the democratization of the certain special effects technologies. And the democratization of the technologies has always had negative effects. Democratizing sound editing tools lead to the rise of pranksters and fake audios, democratizing video recording lead to the appearance of footage taken without consent. In each of the past cases, the net effect of democratization on the World has been positive, and mechanisms for stemming the negative effects have been developed. We believe that the case of neural avatar technology will be no different. Our belief is supported by the ongoing development of tools for fake video detection and face spoof detection alongside with the ongoing shift for privacy and data security in major IT companies.<\/p>\n<hr \/>\n<p>Slightly off-topic, there&#8217;s a new Frontline documentary on AI. It&#8217;s 2-hours, so it&#8217;s a commitment. It doesn&#8217;t really go into the\u00a0GAN-side of artificial intelligence, but it does discuss automation, privacy, and surveillance.<\/p>\n<p><a href=\"https:\/\/www.pbs.org\/wgbh\/frontline\/film\/in-the-age-of-ai\/\" target=\"_blank\" rel=\"noopener noreferrer\"><img loading=\"lazy\" class=\"alignright wp-image-3484 size-full\" src=\"https:\/\/alliwalk.com\/blog\/wp-content\/uploads\/2019\/11\/Screen-Shot-2019-11-24-at-8.03.25-PM-e1574643889663.png\" alt=\"\" width=\"1100\" height=\"505\" \/><\/a><\/p>\n<pre><a href=\"https:\/\/www.pbs.org\/wgbh\/frontline\/film\/in-the-age-of-ai\/\">In the Age of AI<\/a>\u00a0FRONTLINE, from PBS. [1:54:16]<\/pre>\n<p>The documentary provides\u00a0many reasons to be afraid of AI, particularly with regard to surveillance and use of AI by governments.\u00a0We can&#8217;t really predict what governments will do, but\u00a0if behavior control is a goal of AI there&#8217;s a natural user group: people who have trouble controlling their behavior. This would be people who have or have had issues or struggles with:<\/p>\n<ul>\n<li>substance abuse<\/li>\n<li>memory loss<\/li>\n<li>chemical imbalances in the brain<\/li>\n<li>adhd<\/li>\n<li>neurological damage<\/li>\n<li>loss of motor control<\/li>\n<\/ul>\n<p>Or even just reminding people to eat better and go outside more. I&#8217;m sure there are more. Anyway, that&#8217;s my thought. Seems fair to be afraid, but also there are some opportunities that shouldn&#8217;t be overlooked.<\/p>\n<hr \/>\n<p>And now on the cultural side : the trailer for Her. I don&#8217;t know if it&#8217;s possible to have an OS this advanced, but there are some interesting fantasy explorations in this movie.<\/p>\n<p><iframe loading=\"lazy\" width=\"840\" height=\"473\" src=\"https:\/\/www.youtube.com\/embed\/dJTU48_yghs?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture\" allowfullscreen><\/iframe><\/p>\n","protected":false},"excerpt":{"rendered":"<p>A quick overview of generative adversarial networks (GANs) \u2014 a type of artificial intelligence that are capable of generating, among other things, pretty realistic looking photos of humans that do not exist.<\/p>\n","protected":false},"author":1,"featured_media":3416,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[37,1,330],"tags":[555,553,554,541,143],"_links":{"self":[{"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/posts\/3244"}],"collection":[{"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/comments?post=3244"}],"version-history":[{"count":34,"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/posts\/3244\/revisions"}],"predecessor-version":[{"id":3485,"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/posts\/3244\/revisions\/3485"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/media\/3416"}],"wp:attachment":[{"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/media?parent=3244"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/categories?post=3244"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/alliwalk.com\/blog\/wp-json\/wp\/v2\/tags?post=3244"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}