Skip navigation

neural networks

warning: Creating default object from empty value in /var/www/vhosts/ on line 33.

A new poker machine has such smart artificial intelligence that players are hooked even though the house always wins. About 200 machines across the country, called "Texas Hold ‘Em Heads Up Poker," use knowledge gained from billions of staged rounds of poker fed through neural networks, and the result is an unpredictable poker player that can win almost every time. Three different banks of knowledge are used depending on the gameplay scenario, but the basic idea behind its play technique is "to prevent itself from being exploited." "The theory behind it is almost paranoid," as engineer Fredrik Dahl explains. Before the machines hit the casinos, the makers spent two years trying to dumb the AI down so players wouldn't walk away from the machines. Even with the adjustment, it's estimated that only 100 players around the world even have a chance of taking the game down. Michael Kaplan has profiled the machines for The New York Times — be sure to read the full article for all the details.

Your rating: None
Original author: 
Adi Robertson


Stanford Professor Andrew Ng is bringing back the idea of an artificial intelligence that can think like a person. With Google's Deep Learning project, he's creating machines that take a multi-layered approach to information, building up knowledge and figuring out concepts by passing data between various networks that can each recognize a small piece of it. The approach is designed to mimic how the human brain processes information with neural networks, and it's starting to work — last year, Google's "brain" figured out how to identify cats in YouTube videos without being told that the concept of "cat" existed. Wired has profiled Ng and his work on brain-like computers, a project that also ties into current government-funded brain...

Continue reading…

Your rating: None

I'm writing an introductory paper to the field of AI.

What are the major paradigms? What are the major methods?

I'm having a hard time finding a framework for which to approach these topics. It seems to me (a relative outsider) that there are many disparate groups of AI research that overlap while others don't come close at all.

Methods I have so far are neural networks, genetic algorithms, classical (symbolic) AI, expert systems.

Paradigms I would possibly consider are a top-down approach (modelling cognition) and a bottom up approach (genetic algorithms?).

In what way can I group or classify approaches to AI as to help make more sense of this mass? Thank you.

submitted by arohn
[link] [5 comments]

Your rating: None