"On November 19th, according to foreign media reports, like the DIY computer, there are now many artificial intelligent enthusiasts to develop their own artificial intelligence systems using existing tools and data.
In the end of 1975, a message written on the paper was started on the bulletin board near the San Francisco Peninsula. "Are you in your own computer?" It asked like this, "Or any other digital black magic box? If this is the case, you may want to participate in a party."
This invitation attracted 32 people to a garage of California, attended the first party of Homebrew Computer Club, one is the young engineer Steve Wozniak (Steve) Wozniak, he later brought friends named Steve Jobs into the club. The retirement entrepreneur Len Shustek is also in the garage. He said: "This is a proof of personal progress to promote technical progress, and does not have to happen in large companies and universities." "Now, the artificial intelligence area is also happening The same thing."
Since 2012, computers have made significant progress in understanding speech and image, which is attributed to a technique that has been rarely known - artificial neural network. To truly master this artificial intelligence technology, a powerful computer, years of research experience and a deep mathematical basis. If you have all these conditions, congratulations: You are likely to be Amazon, Facebook, Google or other small number of high-pay employees in other technological giants, they compete with extremely complex artificial intelligence strategies to reshape the world.
However, fighting for artificial intelligent hegemon status is also full of various tools and components that can be exposed to anyone. In order to attract top scientists and application developers involved in artificial intelligence, the technology giants released some internal artificial intelligent development kits, as well as related research results. Hackers and hobbies can now use a few close to the same technique to promote the most crazy dreams of Silicon Valley. Artificial intelligence researcher, Entrewen Wu Wenda said: "Today, high school students can do things that are unable to do in the world a few years ago."
People like Wu Weida have great hope for amateur-intelligence of the big outbreak: they hope that it can spread this technology to local and cultural levels to where they stay away from Silicon Valley. Look at the technical parties What will happen when the matter and the world's way to train the "training" neural network. Wu Weida likes to imagine. One day a Indian may take advantage of the video about artificial intelligence to make local drinking water more secure.
Of course, not all homemade (DIY) neural networks can be worried. At the end of last year, some people used the Reddit account to release a porn video, and the heroine seems to be the Gal Gadot in "Wonder Woman". This video is widely circulated in Reddit's overcast corners and adult video websites. However, the careful audience noted that the face of the plus blinks blinks like a loose mask. The final posmer explained that this video is fake, which is to generate a plurality of facial images by training the neural network to match the expression of the stars in the movie. They released code and methods online, so anyone can make similar "deepfake" fake video.
Therefore, it may not be full of sweet and bright, of course, is not full of dark and porn. Most importantly, its expression is very good in terms of specificity. Next, we have inventive a pioneer of DIY artificial intelligence to see what will happen when the public can teach the computer.
I want this neural network to complete the lyrics creation - Robbie Barrat
When Robbie Barrat was in the West Virginia countryside, he began to use the local recycling center to open them, and then re-assemble it. Then he also served in his own farm. When he was in high school, he made a debate on whether the computer had creativity on the computer, and then learned artificial intelligence. The idea of Balat is to use Kanye West's lyrics training neural network learning to sing. Balat likes it at school, but some adults are shocked. Balat said: "The teacher is a bit unhappy because the neural network is too worried."
Illustration: Robbie Barrat hunting time using artificial intelligence design
In the end, the artificial intelligence system full of dirty words was proved to be Barrett left the farm pass. His achievements are not good enough, unable to enter the university in the middle of the university, study math or computer science. But this project helped him find an internship of automobile projects in the center of Silicon Valley. Since then, he moved to Stanford University, now working in a biomedical laboratory, attempting to develop a neural network capable of identifying a pharmaceutical potential molecule. But training neural networks to create art is still his hobby.
Today, in the spare time, Balatte's video clips and photos on the fashion show, training artificial intelligence has produced some models to wearing new clothes. The result is some of the wonders - have you ever thought that you would like a pants that are wrapped around the calf, or a sweater hanging on a huge pocket? However, Balat is working with the designer to turn the results of artificial intelligence into real clothes. He can't wait to try it.
Diagnose plant diseases? This has an app - Shaza MEHDI
Shaza Mehdi's rose bushes in the front yard are very beautiful, but it is easy to get sick. Last day, Mai Hecidi asked himself, why her mobile phone can't diagnose plants in the intensive scientific analyzer in "Star Trek". "How can the computer know?" Asked the high school senior high school students from Lawrenceville, Georgia. Soon, she and a friend called Nile Ravenell, who went to class, manicure and go to the WAFFle House on the school.
Maihedi does not know how to program, there is no such person around her. Adults in her life can encourage her to do this, but they can't provide professional knowledge; she does not open the introduction of computer science. Maihedi lying in bed, Teddy, Teddy, and her old Dell laptop together, self-study by Youtube video and online tutorial, based on the basics of Python and neural network basics. When she encountered BUG in the program, she would help to the strangers on the forum. "I really hate bug," she was happy to recall.
Diagram: Programming "Xiaobai" Maihedi developed artificial intelligence to diagnose plant diseases
It is worth mentioning that Maihedi is inspired by a YouTube video, its content is the neural network developed by a researcher at Stanford University to identify skin cancer, and the effect can be compared with professional dermatologists. A online tutorial tells her how to achieve the skill of neural network alone. The first step is to download the software to identify daily necessities, such as toilets and teapots. The second step is to re-adjust its vision, give it approximately 10,000 tagged diseases, these pictures are Mahdi collected from the Internet, and have been identified by disease.
In 2017, she finally tested her application Plantmd. Maihedi looked at the application to scan a leaf with a loud and brown spots, and then a sparked leaf suddenly appeared on the mobile phone screen. After a few times of tense heart skip, the word "vine anthracnose" was flashed from above. Fast network search confirmed this diagnosis: a clear case of a fungi infection, also known as bird eye rot. "I put down my heart," Maihedi recalls, "" Handheld scientific recorder work. "
The camera can check your clothes - Tianyuan Dahara (Daisuke Tahara)
In the small city of Japan, dry cleaning is a daunting job. The family of Daisuke Tahara has 8 dry shops in Tachae. Takawa is located in the south of Japan, with a population of only about 50,000. Therefore, Tianyuan began to consider allowing the computer to share some work.
First, 38-year-old Tianyuan tried to achieve business modern, better record and track orders through a better computer system. But most of his employees have lack in technical experience, they are difficult to adapt to this system. "They are easy to forget," Tianyuan said. So, this self-study programmer began research software to automatically check the customer's clothes. He read books on machine learning and played its English and programmed skills to the extreme. In his own store, he took 40,000 photos about a suit, shirt, skirt and other clothes, and used them to train their own artificial intelligence.
In July this year, Tianyuan began to test the system in one store. The customers put their clothes on the table, and the head is the camera. Tianyuan's software looked at it and showed its conclusions on the tablet (two shirts, one jacket). Employees usually need to help customers in the first time. After that, the customer who came to dry the store can use it alone.
Tianyuan said that his workers had doubts about his inventions, but after discovering that software makes their work easier, change the original view. He didn't plan to use this project as an excuse of layoffs, but he hopes to help expand its business. He said: "I want to open a store with this system, there is no shop."
Waymo's epitome - Will Roscoe
In a warehouse in Auckland, California, a small group of books are watching Will Roscoe with thumb click on the phone. At his feet, a plastic shell is torn with a remote control model to drive along the yellow and white tape labeled runway on a scratched concrete floor, and Rosco does not control. This car named "Science Monsters" has a camera with a bunch of electronic devices at the top, known as "donkey" donkey car. Rosco is not an artificial intelligence expert, but his work is similar to the way Waymo automatic cars perceived world.
Dialect: Will Roscoe invented the "Truck" Donkey CAR platform
As a well-trained civil engineer, Rosco designed Donkey Car because of the election experience. In 2016, he campaign in the subway system of San Francisco Bay Area (Bay Area), Rosco promised to use the automatic driving electric bus to replace the existing subway to expand its capacity, but he finally only won the third place. It is a good way to show your own small automatic driving cars seem to show this technology to voters. He said: "I want to prove that it can play in a small range."
It turns out that his time is very perfect - a robotic fan group specially cracking the remote model car is about to hold the first meeting in the nearby Berkeley. There, he encountered another repairman Adam Conway, and he proposed such a car. Self-learning programmers Rosco use Google developed open source software Tensorflow created their own automatic driving system. He also borrowed some neural network code from a participant from the remote model car party. The final design of Roscoo is to learn driving through the observer driving a car during the demo. He named the model car created as "dollar", because he believes that the scorpion is safe, not an elegant in the traditional sense, and it is easy to experience the situation.
Rosco and Conway put all their software and hardware design to users for others. "Cars" enthusiasts are now playing in Melbourne in Hong Kong, Paris and Australia. In January of this year, in the reputation in Auckland, nine homemade automatic driving car competitions run a single-zone track in the fastest speed; there is a "dollar" manufactured by three tension high school students in the competitor. These vehicles have also begun to come out of the game. Two hobbies near Los Angeles have transformed a car to find and pick garbage on the beach.
Original address: https://www.eeboard.com/news/diyer-ai/
Search for the panel network, pay attention, daily update development board, intelligent hardware, open source hardware, activity and other information can make you master. Recommended attention!
[WeChat scanning picture can be paid directly] "
Our other product:
