Online: | |
Visits: | |
Stories: |
Brett Smith for redOrbit.com – Your Universe Online
Engineers at the University of California, Santa Cruz have taken a bold step forward with the development of a cutting-edge artificial intelligence circuit.
According to a new report in the journal Nature, the UC Santa Cruz team was able to use the artificial circuit to successfully perform a relatively complex task we do every day: classify images.
Our ability to classify images goes beyond just being able to organize the digital photos on your hard drive. In fact, you’re classifying the images or letters and words right now as you read this sentence. As you read, your brain sorts through the visual cues on the screen and processes them based on the surrounding context of the entire article and the redOrbit website. All of this is processed within fractions of a second.
Using their novel artificial intelligence circuitry, the study team was able to classify three letters – “z”, “v” and “n”—by their images. The system even held up when each letter was highly stylized or flooded with “visual noise.” In a method similar to how we find the correct key from a ring of similar keys, the AI neural circuitry was able to properly classify the simple images, according to the new report.
“While the circuit was very small compared to practical networks, it is big enough to prove the concept of practicality,” said study author Farnood Merrikh-Bayat, a post-doctoral computer engineer at UC Santa Cruz. “And, as more solutions to the technological challenges are proposed, the technology will be able to make it to the market sooner.”
Now that’s classified information!
The AI circuit is based on a device called memristor, the portmanteau of “memory” and “resistor.” The novel device has an electronic element whose amount of resistance shifts depending on the direction of the flow of the electrical charge. Unlike customary transistors, which depend on the drift and diffusion of electrons and their holes through semiconducting material, memristor functioning is by using ionic movement, comparable to the way human neural cells produce neural electrical signals.
Study author Dmitri Strukov said the ionic nature of the device has many advantages over purely electron-based models, making it very desirable for artificial neural network use.
“For example, many different configurations of ionic profiles result in a continuum of memory states and hence analog memory functionality,” said Strukov, an assistant professor of computer engineering. “Ions are also much heavier than electrons and do not tunnel easily, which permits aggressive scaling of memristors without sacrificing analog properties.”
The researchers said many more memristors would be necessary to build more advanced neural networks to do the same types of things we can do with hardly any effort and energy, such as identify several versions of the same object.
—–
Follow redOrbit on Twitter, Facebook, Google+, Instagram and Pinterest.
redOrbit.com
offers Science, Space, Technology, Health news, videos, images and
reference information. For the latest science news, space news,
technology news, health news visit redOrbit.com frequently. Learn
something new every day.”