Deep Learning Machine Solves the Cocktail Party Problem

Everyone is familiar with the cocktail party effect. When chatting in a crowded space, humans can focus on a particular speaker’s voice, while filtering out other voices and noise.

We do this without really thinking about it, effortlessly. But it turns out to be a surprisingly difficult problem to solve for a machine.

However, a team at the University of Sussex have developed a technique to separate a singer’s voice from the music in songs. Karaoke anyone?

Deep Learning Machine Solves the Cocktail Party Problem | MIT Technology Review.

Advertisement

Alien First Contact Could Be With An Artificial Intelligence

Arguably, the most successful robots to date have been our interplanetary explorers. The twin Mars Exploration Rovers, Spirit and Opportunity far exceeded their operational design time, lasting years instead of weeks.

Voyager 2, launched in 1977, has recently left the solar system, becoming the first man-made object to do so.

V’Ger from Star Trek: The Motion Picture

It’s reasonable to assume that our first interstellar explorer will be followed by many more. Indeed, the first explorer from Earth to reach another planet may well indeed be an AI-powered robot, albeit far in the future. This is far more likely than a human astronaut making the journey.

As noted in the Daily Express article below, this implies that our first contact with extraterrestrials may well be with an Artificial Intelligence. The article is based on a recent paper by Susan Schneider, who suggests that we should be looking for signals from artificial life. Such life may not be dependent on the current requirements for organic life and we could widen the scope of programs like SETI.

It’s also likely that should ET show up at our planet, it would be an AI. Given the recent negative stories on AI, this may not be a good thing. And we don’t even have the Avengers to look after us.

All over for God: Dominant life force in universe is ‘super-intelligent alien space robot’ | World | News | Daily Express.

Autoencoders explained in one page

Will our smartphones be more intelligent than us?

A somewhat alarmist article appeared in The Canberra Times yesterday (April 26th 2015). This particular excerpt is quite speculative:

Within seven years – about when the iPhone 11 is likely to be released – the smartphones in our pockets will be as computationally intelligent as we are. It doesn’t stop there, though. These devices will continue to advance, exponentially, until they exceed the combined intelligence of the human race.

The analogy here though is quite flawed. An iPhone (indeed, any mobile phone) is simply mobile computer hardware. It is the software that makes it actually do things, such as making calls, surfing the internet and running applications.

So although the phone may have a large amount of computational power, it will be advances in software that will largely determine how it behaves running AI software.

Today’s research is focused on building smart algorithms that solve specific problems, not into building generally intelligent machines. As a result, the futuristic iPhone is more likely to have dozens of AIs rather than a single one, each AI very good at one thing, but completely clueless at others.

Of course, this also depends on rapid advances in battery technology. Otherwise the phone may die before the AI gets to “Hello”.

via The coming problem of smartphones being more intelligent than us.

Carnegie Mellon Computer Faces Poker Pros in Epic No-Limit Texas Hold’Em Competition

One of the main areas of Artificial Intelligence research has been into game playing. In many games such as chess and draughts, there is perfect information. Each move a player makes is seen by the other player.

Some games have elements of chance, in addition to perfect information. Backgammon is a good example, where each move a player makes is governed by the roll of a dice. Still, there is perfect information available as to the state of the board.

Texas Hold’Em Poker – Todd Klassy

Poker is an interesting game by contrast as there is uncertainty about the state of the game. The only things a play knows for certain are his or her own cards (and in the case of Texas Hold ‘Em, the shared cards). Everything else must be estimated using some element of probability. In the case of poker, it would be what cards the other players have and the probability of the cards to be dealt by the dealer during the betting rounds.

There are various techniques and strategies employed by human players, including reading the body language of the other players. This is probably beyond today’s level of AI, but that hasn’t stopped a team from Carnegie Mellon University from having a go. From April 24th until the 8th May, an AI will be playing a heads-up no-limit Texas Hold ‘Em against professional players. It will be interesting to see if it can follow in the footsteps of Deep Blue, the AI that defeated the then world chess champion, Gary Kasparov, in 1996.

For the latest score in the match, see Brains vs AI at Carnegie Mellon

Sentient Artificial Intelligence, yay or nay?

theblackpeppersauce

SENTIENCE 

---

is the ability to feel, perceive, and experience subjectively.

– wikipedia

Scientists and philosophers have been pondering if we will ever create artificial intelligence that will rival human intelligence. Robots that are considered man-made machines that are just as intelligent, creative, and self-aware as its creator.

While reasons backed with argumentative research and scientific explanations are all over the internet, I wanted to see more of the ‘back-at-home-behind-my-PC’ perspective.

I asked two “netizens” to share their thoughts about this matter. One was on the affirmative and the other was against it. Here’s what they had to say. On a positive light,

It is a huge step forward.

– T.Sorrera

The first response stated that sentient AIs will replace jobs such as repetitive work, household chores, and delivery, to toxic waste disposal, one-way trips to outer space, and security. Since AIs are inevitably more accurate calculators compared to human brains…

View original post 169 more words

Google’s Schmidt On NSA Snooping, Drones, Artificial Intelligence And More | Androidheadlines.com

Quote from Eric Schmidt on Artificial Intelligence and autonomous cars.

Talking about something that has hitherto strictly been in the domain of science-fiction, but is slowly becoming a reality, Mr. Schmidt asserted his belief, that the use of AI (Artificial Intelligence) makes people more productive. That’s where Google’s much talked about experiments with drones and self-driving cars come in. He pointed out that Google only “added software and navigational capability and particularly high quality laser which runs on the top of the car. We’re actually able to watch what’s going on better than you are. The laser sees better than the human eye”.

Google’s Schmidt On NSA Snooping, Drones, Artificial Intelligence And More | Androidheadlines.com.

A self-driving car to die for: The Chevrolet-FNR

Now this is work of art! Chevrolet’s autonomous electric car is straight out of the Tron: Legacy movie. With it’s sleek lines and electric blue colours, it’s certain to be a hit.

Sadly though, like most concept cars, it’s very unlikely to appear at  your local showroom.

Chevrolet’s autonomous electric concept car has a swivel chair driver’s seat | Ars Technica.

Swordfighting robot leads to more irrational AI fear

Any article in the popular press about Artificial Intelligence that includes a picture of the Terminator robot is guaranteed to be focused on one thing alone, building up the fear.

Source: News Corp Australia

According to the article below, Japanese scientists have created a robot that can swordfight. This naturally leads to the suggestion that we are on the brink of having Terminator robots running all over the planet killing at will.

While it makes for a good story, it largely ignores the fact that this robot probably can’t do very much else. This trait is shared by most AI algorithms: they are brittle. This isn’t necessarily a bad thing. Because it is focused on a particular area, an AI can do a lot of good in that area. It’s lack of use in another area lends itself to being controlled. This Artificial Narrow Intelligence is what we have today and is the state of the art.

Terminator robots are examples of Artificial General Intelligence, and we are a very long way away from that.

Scientists create a swordfighting robot, ethicist warn of the danger of artificial intelligence.

The robots are coming, and they want your job

This is a very good article on some of the potential impacts of mass production of robots and Artificial Intelligent machines. There are going to be huge impacts on society, economics and wealth generation.

That means any job a human can do will soon be done by a machine which will probably do it even better. Plus, they won’t require rest, they’re more easily replaced and upgraded, and they don’t demand higher wages or join pesky unions.

Scary stuff indeed.

There is going to be a lot of regulation required on the use of such machines, especially when wealth generation will be concentrated in the hands of a few.

This will happen long before the big philosophical questions arise about intelligence, sentience and rights for machines. The main reason is that Artificial Intelligence won’t be one all-encompassing intelligence. Instead it will be a thousand different AI’s, each one focused on being superior in one area or on one class of problems.

In any event, how long will it be before protests start and politicians react? Perhaps in the next decade or so. All the signs point to rapid industry adoption of AI technology which can only mean less work for humans.

via The robots are coming, and they want your job.