SoylentNews is people
SoylentNews
Using the ImageNet object classification benchmark, Baidu’s Minwa supercomputer scanned more than 1 million images and taught itself to sort them into about 1000 categories and achieved an image identification error rate of just 4.58 percent, beating humans, Microsoft and Google. Baidu's Minwa scored 95.42%, Google's system scored a 95.2%, and Microsoft's, a 95.06%, Baidu said.
“Our company is now leading the race in computer intelligence,” said Ren Wu, a Baidu scientist working on the project. “I think this is the fastest supercomputer dedicated to deep learning,” he said. “We have great power in our hands—much greater than our competitors.
A paper released Monday [May 11, 2015] is intended to provide a taste of what Minwa’s extra oomph can do. It describes how the supercomputer was used to train a neural network that set a new record on a standard benchmark for image-recognition software. The ImageNet Classification Challenge, as it is called, involves training software on a collection of 1.5 million labeled images in 1,000 different categories, and then asking that software to use what it learned to label 100,000 images it has not seen before.
(Score: 1, Funny) by Anonymous Coward on Saturday May 16 2015, @11:05PM
Because all Asian faces look the same, Baidu has to work extra hard to tell them apart.
(Score: 0) by Anonymous Coward on Saturday May 16 2015, @11:35PM
Americans are two-faced, so the Microsoft and Google algorithms had to work twice as hard.
(Score: 2) by looorg on Saturday May 16 2015, @11:45PM
If they all look the same how would we know if it even works? They could just be an endless stream of false positives.
(Score: 2, Interesting) by Anonymous Coward on Sunday May 17 2015, @12:20AM
While I know you mean that as a joke, to give a serious answer you can take a picture and label it at the time of taking it. Take another picture and see if the software can match the two together when given multiple pictures.