}#PageList1 {margin-bottom:0px} .content-outer { -webkit-box-shadow:none; box-shadow:none; } #ContactForm1 { display: none ! important; } .main-nav { width: 100%; position: absolute; left: 0; margin-top: -45px; } #header { margin-top: 45px; }.content-inner { margin-top: -60px !important; } -->

October 22, 2012

Marcus Hutter On Universal Artificial Intelligence

Marcus Hutter

 Artificial Intelligence
In his 2008 book Universal Artificial Intelligence, Marcus Hutter rigorously attempted to define artificial intelligence, and formulate a strategy to produce artificial general intelligence systems.  Since then he has continued to develop what he calls Universal General Intelligence (AIXI) and foresees the time when it can exceed human capabilities.
Marcus Hutter, author of Universal Artificial Intelligence: Sequential Decisions Based on Algorithmic Probability, has a complex and intertwined theory of intelligence and the possibilities for Artificial General Intelligence (AGI).

In 2000 he joined J├╝rgen Schmidhuber's group at the Swiss Artificial Intelligence lab IDSIA, where he developed the first mathematical theory of optimal Universal Artificial Intelligence (AIXI), based on Kolmogorov complexity and Ray Solomonoff's theory of universal inductive inference. In 2006 he also accepted a professorship at the Australian National University in Canberra.

Hutter's notion of universal AI describes the optimal strategy of an agent that wants to maximize its future expected reward in some unknown dynamic environment, up to some fixed future horizon. This is the general reinforcement learning problem. Solomonoff/Hutter's only assumption is that the reactions of the environment in response to the agent's actions follow some unknown but computable probability distribution.

According to Hutter,
[Once] we have [an] informal definition that intelligence is an agents ability to succeed or achieve goals in a wide range of environments. The point is you can formalize this theory, and we have done that and it is called AIXI. Or Universal AI is the general field theory and AIXI is the particular agent which acts optimally in this sense. 
So that works as follows: it has a planning component, and it has a learning component. What the learning component does is, think about a robot walking around in the environment, and at the beginning it has no data/knowledge about the world, so what it has to do is acquire data/knowledge of the world and then build its own model of the world, how the world works. And it does that, so there are very powerful general theories on how to learn a model from data, from very complex scenarios.

In order to acheive outcomes for the artificial intelligence system that were based on the data-sets provided, Hutter turned to a foundation of scientific reasoning - Occam's Razor.  By formulating algorithms that analyze data and produce models of it based on the simplest explanation, Hutter was able with AIXI to achieve a working form of artificial general intelligence.

To date, this system has been employed in simple or "toy-like" applications, with the system playing video games, however Hutter insists that AIXI can eventually be scaled to meet and exceed human capabilies in the not-to-distant future.

According to Hutter, the approaches to artificial intelligence (AI) in the last century may be labelled as: (a) trying to understand and copy (human) nature, (b) being based on heuristic considerations, (c) being formal but from the outset (provably) limited, (d) being (mere) frameworks that leave crucial aspects unspecified. This decade has spawned the first theory of AI, which (e) is principled, formal, complete, and general.

Marcus Hutter

Hutter's theory, called Universal AI, is about ultimate super-intelligence. It can serve as a gold standard for
General AI, and implicitly proposes a formal definition of machine intelligence.

In the talk embedded below, Hutter provides, after a brief review of the various approaches to (general) AI, an introduction to Universal AI, concentrating on the philosophical, mathematical, and computational aspects behind it. He also discusses various implications and future challenges.

SOURCE  Adam Ford

By 33rd SquareSubscribe to 33rd Square

No comments :

Post a Comment

Thank you for your comment!