Microsoft: Computers that Outperform Humans
Microsoft claims to have developed a computer system capable of outperforming humans at classifying images through a popular dataset called ImageNet. It is estimated that currently humans can classify images on ImageNet with an error rate of 5.1 percent, while Microsoft’s deep-learning system had an error rate of 4.94 percent.
Microsoft released its error rate after Baidu (A popular browsing system located in China) claimed to have achieved an error rate of 5.98 percent. This was huge, because Baidu’s record setting system was achieved using homemade supercomputing architecture. However, in an official competition hosted on ImageNet, the best performance went to Google. Their error rate in 2014 was 6.66 percent.
According to Image-Net.org, “ImageNet is an image database organized according to the WordNet hierarchy (currently only the nouns), in which each node of the hierarchy is depicted by hundreds and thousands of images…We hope ImageNet will become a useful resource for researchers, educators, students and all of you who share our passion for pictures.”
In a Microsoft Blog, researcher Jian Sun explains their findings, and the difference between the computer systems they developed and human performance.
“Humans have no trouble distinguishing between a sheep and a cow. But computers are not perfect with these simple tasks. However, when it comes to distinguishing between different breeds of sheep, this is where computers outperform humans. The computer can be trained to look at the detail, texture, shape, and context of the image and see distinctions that can’t be observed by humans.”
Mr. Sun is saying that the main difference between MicrosoftThis observation is very interesting and important to Microsoft’s research; it tells us that while computers are capable of recognizing many factors of an image, they are not perfect. Humans are more effective at recognizing simple effects of an image but they are also not perfect.
This technology will be useful to people because it has many applications. It can be used by the proper authorities to recognize people. It can also have an application on motor vehicles to help people drive safer. It might even be used in space, to study foreign material located in other galaxies.
If you would like to find out more about artificial intelligence, there are a number of conferences happening this year (2015). One is called Structure Data, in New York on March 18th and 19th. Another conference called Structural Intelligence, is on September 22nd and 23rd in San Francisco.
Scheduled to speak at these conferences include deep learning and machine learning experts from Facebook, Yahoo, Microsoft, Spotify, Hampton Creek, Stanford and NASA, as well as startups Blue River Technology, Enlitic, MetaMind and TeraDeep. Mark your calendars and attend these revolutionary conferences that will enlighten you with research done by top players in this industry.
Harris, Derrick. “Microsoft Says Its New Computer Vision System Can Outperform Humans.” Yahoo Finance. GIGAOM, 13 Feb. 2015. Web. 16 Feb. 2015.