Lei feng's network (search for "Lei feng's network", public interest) by: this article is translated from the Google's by Pu technology engineer Artificial Brain Is Pumping Out of the Trippy-And Pricey-Art. Web exclusive article of Lei feng.
On Friday night, in the San Francisco Mission District (Mission District) of a vintage movie art gallery, Google graphics expert Blaise Ag • era y Arcas to approximately 800 geek hipster published a wonderful speech.
Used to hang a film his speech when he stood on the wall next to the screen, projection equipment used in this piece displayed 500 years ago on the walls of Renaissance Germany a pair of double portrait of the painter Hans Holbein paintings. Paintings in the distortion of a human skull, according to Blaise's interpretation of Hans Holbin was unlikely to directly draw the odd human skull, but use mirrors or lenses to project images of the skull to the canvas, and then sketch out the contours of the skull. "He is using the most advanced technology at that time," Blaise said.
Through this example, Blaise mean we use science and technology to create art has a history of several centuries, at this point, and now we're not so different from past. Blaise Gallery through this way each introduce the exhibits created by artificial neural network, which consists of computer hardware and software combination of neural network and neural network similar to the human brain. Last year, Google researchers have developed a new kind of artificial neural network for artistic creation, this weekend, Google put the artificial intelligence system into a two-day exhibition which generates a picture for a commitment to the arts and technology and the nonprofit organization--the gray area Foundation for the Arts (the Gray Area Foundation for the Arts)-had raised $ 84,000. cover Moschino
Night of the exhibition, is a particularly stylish but geeky Valley scene. "Look at it! This is Clay Bavor, Google's famous head of the virtual reality project! "。 "Look, there is a TechCrunch (a tech media) by Josh Constine. "" This is MG Siegler, writes for TechCrunch in the past, now often appear in artificial neural network art exhibitions, well, I think it was him. "At the same time that night and demonstrated a fast pace of development of artificial intelligence technology. Artificial intelligence has now reached a historic high, artificial neural networks can not only promote the progress of the Google search engine, but also produces people willing to pay big money to buy the works of art.
In the eyes of Blaise, from mankind's first since the artwork was created, these are merely an integral part in the process of history, just like Han Holbein's neural network just broadens the creative form of art. While in others, this is an exciting new things. "This is the first time I viewed the creation of works of art into a science project. "One gray area Foundation for the Arts patron Alexander Lloyd said he spent thousands of dollars on a pair of artificial neural network's works of art. Friday night's show at the same time has given us a positive hints, make us more inclined we will have a wonderful world of artificial intelligence, smarter machines will become more independent, they will engage in more work for us and led us to a new world beyond our imagination.
| Depth (learning) dream
Today, including Google, Facebook, Twitter, Enterprise, and artificial neural networks have been widely used in automatic recognition of images, the smart phone's voice command recognition and intelligent translation between different languages. If you have entered enough pictures of your family to the machine, then the machine will learn to recognize your kin. This is Facebook the way you upload photo in identify face. Now, called "deep dream (DeepDream)", "generator", Google successfully subvert these neural networks. Is no longer recognized the picture of the machine, but to create the picture.
Google said the results called "Inceptionism", to the movie inception 2010 Leonardo (Inception), pay tribute to, the film depicts a can im going to put our ideas into the consciousness of the other technologies. Although this may not be the best analogy, but this new technology does is it shows us robot dreams.
In order to train "deep dream", "brain", we first need to enter a photo or other image, artificial neural networks will be looking for similar patterns in these pictures, constant repetition will strengthen this pattern. "This would create a feedback loop: If a cloud looks a little like a bird, artificial neural networks will make it look more like a bird," staff at Google in a blog post announcing the project, said for the first time. "This will, in turn, makes artificial neural network better able to identify birds in the next step, until a rich detail bird suddenly appear in front of you. " iPhone 6 Moschino
Output results in this way is very charming, but a bit disturbing. If you put your photos into artificial neural networks, if it found that somewhere there are some on your face like a dog, it will face this part of your dog. "This is like a neural network was hallucinating," said Steven Hansen, a Google DeepMind AI intern of London lab, "it looks like dogs everywhere! "Or, if you give an artificial neural network input random noise picture, it might generate a tree or a tower, or a whole consisting of Tower City. In the same amount of noise in the picture, it may find a pig and a snail's blurry images, then it will generate a frightening new creatures that combines both characteristics. Think about it, it's like a machine taking hallucinogens.
| Real art (Virtually Art)
For developers of Google engineer Alexander Mordvintsev, vision is initially developed this technology to better understand the behavior of neural networks. Although the artificial neural networks has to show its power, but in the way they behave, there are still some mysteries have been solved. We cannot fully understand this hardware and software combination of what happened to the neural network. Mordvintsev and other researchers were still working on research of artificial neural network in order to fully understand them. At the same time, another engineer at Google, Mike Tyka went to look for the ability to use this technology to create art. His research on artificial neural network for Google at the same time, was also a sculptor. He saw the technology as technology combined with the art of a way.
Artists with Tyka, responsible for selected image input to artificial neural networks, they can to adjust certain behaviors of artificial neural network to the appropriate mode. They can even start training them so that they recognize the new pattern, in order to release seemingly limitless possibilities. This technology generates some of the works look very similar, especially in a number of works about spiral, dogs and trees. But part of the work in their own adventures on the direction, showing a through inhospitable and cold machines picture.
On Friday night, Tykad artificial neural networks responsible for four pieces of art created by auction. Are the "drill Castle Air Museum (Castles in the Sky With Diamonds)", "devout faith of God's Army (Ground Still State of God's Original Brigade)," "the whim of the Carboniferous (Carboniferous Fantasy)", "Sparks, sun shining Babylon (Babylon of the Blue Sun)". Through the Gallery, the odd painting, a match the singular visual effect of paintings. Not surprisingly these unique paintings, according to the plan the exhibition host Joshua said, these names are these works of art of artificial neural networks of their choice. Ross Goodwin, one of the New York University graduate student used the same type of artificial neural network technology to generate names for Tyka work.
In the view of Steven Hansen, these paintings that are automatically generated and cannot be regarded as a major leap forward. "I felt like an advanced version of PhotoShop," he said. But at least the DeepDream symbolizes the great changes in this area, the machine can already do a lot of things for myself. On the Google search engine, which changes all the more obvious. Machines did more work at the same time, means we humans play a role in search engine will become less, or at least we have decided on the right of the search results more and more far.
Moreover, this gap not only in the Google search engine, but spans many fields of the online services and technologies. On Friday, in a corner of the Gallery, Google official invite visitors to wear their paper VR equipment to explore "deep dream (DeepDream)" deep inside. Cardboard VR equipment to show people a realistic "simulated universes" a little. Moreover, the technology is undergoing rapid development period. So it is no exaggeration to predict that one day, machines will be able to create a much larger virtual world. Google VR Clay Bavor, head of the project and not just as a visitor appears in the exhibition, he was behind the weekend Exhibition sponsor and sponsor one of Joshua To work in Google VR.
As you can see, Hans Holbien using technology to accomplish their art, and this line of thinking will radically change some areas, causing profound changes.
Lei Feng network Note: reprints please contact us for authorization and keep the source and author, no deletion of content.
No comments:
Post a Comment