DeepMind: un paso más hacia el desarrollo de computadores que piensan – El Mostrador

Los científicos que trabajan en esta unidad de inteligencia artificial que fue adquirida hace dos años por Google, consiguieron desarrollar un DNC, que consiste en un computador neuronal diferenciable (DNC) capaz de resolver problemas a pequeña escala sin conocimientos previos.

Fuente: DeepMind: un paso más hacia el desarrollo de computadores que piensan – El Mostrador

Google hires leading quantum computing expert –

Google hires leading quantum computing expert –

Google today announced that it is expanding its research around quantum computing and that it has hired UC Santa Barbara’s (UCSB) John Martinis and his team – one of the most prolific research groups in this area — to work on new quantum processors based on superconducting electronics.

Google has hired one of the world’s leading quantum computing researchers as it ramps up efforts to develop artificial intelligence and vastly increase the processing power of computers.

Physicist John Martinis and his team at the University of California Santa Barbara will join Google’s Quantum Artificial Intelligence Lab, a collaborative project between the technology group, Nasa and the Universities Space Research Association, a non-profit organisation that studies space. The team will form part of an effort “to design and build new quantum information processors based on superconducting electronics”, said Hartmut Neven, Google’s director of engineering, in a blog post.




It is the latest sign of Google’s bet on the rise of “smart machines”, which it is developing along with a host of experimental projects from drones to self-driving cars.

In January, Google paid £400m for UK start-up DeepMind, whose brain-like “neural network” algorithms can be set loose on huge data-sets and learn in a similar way to the human mind. Its technology is particularly good at calculations based on pattern recognition, such as image searching or looking for the cheapest or best route to a destination.

Google is already working with Nasa to develop applications on a D-Wave quantum computer, the only quantum device that is commercially available, although there is dispute as to the extent to which it is a genuinely “quantum” device.

Developing such technology could help run the sophisticated algorithms that would be required to develop “intelligent” machines, experts say.

Mr Martinis explained the potential reach of quantum technology in a presentation to Google last October. “It’s a physics nightmare . . . We’ve been going at it for 20 years,” he admitted.

Although his team has not yet built a full computer, they have shown how it is theoretically possible to use electrons’ unique ability to exist in two atomic states to vastly increase computing power, because it allows multiple calculations to be run through the system at the same time.

Anders Sandberg, a computational neuroscientist at Oxford university’s Future of Humanity Institute, said quantum technology is likely to be useful for running sophisticated search algorithms for unordered data. Much of what is on the web falls into this category and more is likely to be produced by the rise of connected devices and the “internet of things”.

In depth



Google is showing results from heavy investments in areas beyond search, with notable inroads in the mobile, video and display markets

“The interesting thing about quantum superpositions is that you cannot just do several things at once, but you can tease out patterns in clever ways,” he said. “[Quantum computing] is very cutting edge and we don’t know if it’s going to work out – but there are tantalising hints that it could.

“A lot of artificial intelligence is about searching for patterns and connecting stuff,” he added. “Our brains are using really slow neurons, but using them really well because they’re running in parallel.”

For example, Prof Martinis has said it would take a bank of computers the size of North America running for 10 years and consuming the earth’s entire store of energy every day to figure out all the prime numbers contained in a 2,000-long sequence of binary code. A quantum computer the size of a lecture theatre could do the same calculation in a day.

Mr Sandberg said factoring numbers can be useful for encryption and code-breaking by governments.

But while versions of quantum computers appear to be capable of doing specific, focused applications fast, making them versatile has been a challenge.

“The fact you can have big arguments about whether [the D-Wave] is a quantum computer shows it’s early days,” Mr Sandberg said.