Posted by ljmacphee on November 30, 2007 under artificial intelligence in the news | 
Lightness illusions are fundamental to human perception, and yet why we see them is still the focus of much research. Here we address the question by modelling not human physiology or perception directly as is typically the case but our natural visual world and the need for robust behaviour. Artificial neural networks were trained to predict the reflectance of surfaces in a synthetic ecology consisting of 3-D “dead-leaves” scenes under non-uniform illumination. The networks learned to solve this task accurately and robustly given only ambiguous sense data. In addition—and as a direct consequence of their experience—the networks also made systematic “errors” in their behaviour commensurate with human illusions, which includes brightness contrast and assimilation—although assimilation (specifically White’s illusion) only emerged when the virtual ecology included 3-D, as opposed to 2-D scenes. Subtle variations in these illusions, also found in human perception, were observed, such as the asymmetry of brightness contrast. These data suggest that “illusions” arise in humans because (i) natural stimuli are ambiguous, and (ii) this ambiguity is resolved empirically by encoding the statistical relationship between images and scenes in past visual experience. Since resolving stimulus ambiguity is a challenge faced by all visual systems, a corollary of these findings is that human illusions must be experienced by all visual animals regardless of their particular neural machinery. The data also provide a more formal definition of illusion: the condition in which the true source of a stimulus differs from what is its most likely (and thus perceived) source. As such, illusions are not fundamentally different from non-illusory percepts, all being direct manifestations of the statistical relationship between images and scenes. [ read more Cognitive Daily: Artificial networks see illusions too]
More information:
Auntie Em’s House of Cookies
What are lightness illusions and why do we see them?
White’s illusion
See also:
Computer Model mimics brain and processes visual information
Posted by ljmacphee on November 28, 2007 under artificial intelligence in the news, bots |
I’m not sure which is worse? The fact you can’t escape election propaganda even in your email in box anymore? Or that the government has found yet another way to intrude into our lives using the internet, artificial intelligence and datamining?
Allegations have been made that Ron Paul’s campaign is creating internet buzz with spam sent through illegal botnets. I’m sure the rest of the candidates won’t be far behind. What does it say about a candidate that he is willing to use criminal methods to help his chance of getting elected? Last time that happened Richard Nixon won the White House. Right now candidates do polls and tell various political factions what they want to hear. How long before those speeches are geared specifically to you and arrive in your in box?
More information:
Analysis: is Ron Paul internet buzz real or spam?
Criminal botnet stumps for Ron Paul, Researchers Allege
First Presidential Candidate spam
Posted by ljmacphee on November 26, 2007 under artificial intelligence in the news |
In order to understand AT&Ts complicity in the recent surveillance with out warrants scandal you might look to ‘The Hacker Crackdown, Law and Disorder on the Electronic Frontier’.
AT&T was an early victim of crackers and one of the first companies building defenses. AT&T being a very old school company was a little unclear on the concept and has certainly gone above and beyond the call of duty in fighting crackers. One of the methods developed to deal with phone crackers was a large scale program to track crackers and their associates and to look for people calling the same circle of friends a someone previously banned from the AT&T network.
A brilliant idea, Hancock is implemented in C. Not only does is provide for building the graph of connections but it allows you to see how the graph changes over time. It sounds promising yet I could find no published case studies showing its effectiveness. Instead of worrying about handling the large volume of data you can concentrate on what you wish to pull out of the data.
Then of course there are the legal and social aspects of all this data mining. Was AT&T legally collecting data for the government? What is the government doing with the data and with people it finds closely connected to terrorists? Does the software work? Are we ruining innocent people’s lives or are we saving innocents from terrorists? There is much yet that needs to come to light about this program. The government is using it as a ‘guilt by association’ program.
More information:
FBI data mining reached beyond initial targets
AT&T Invents Surveillance Programming Language ( Slashdot )
AT&T Invents Surveillance Programming Language ( Wired )
Communities of Interest, (pdf)
Hancock: A Language for Processing Very Large Scale Data
Method of inferring behavioral characteristics based on a large volume of data
Source Code and binaries for Hancock
Manual for Hancock( pdf )
The Hacker Crackdown, Law and Disorder on the Electronic Frontier
See also:
Software recognizes short and long term anxiety in people
Big brother arrives via Comcast 24 years late
Posted by ljmacphee on November 23, 2007 under artificial intelligence in the news, bots |
I wonder if we could get one of these to run for president? It’s bound to be progress.
The two year old Artificial Intelligence (AI) known as the Buddhabot began answering questions on Yahoo! Answers site last week. Yahoo Answers is a Web 2.0 site with a social content rating system reminiscent of Digg. The Buddhabot has so far answered 102 questions and eleven have been selected as the best answer. The Buddhabot is the first and only AI to compete with human beings to provide the best answers on Yahoo Answers new social networking site. . .
Ingram says that if the Buddhabot can demonstrate even an average score on Yahoo! Answers this is tantamount to passing a variant of the Turing Test, a test proposed in the 1950’s by the famous British Scientist Alan Turing to prove computer consciousness. Turing suggested that if a machine could convince a human being that they were talking to another human instead of a machine that the machine might be considered intelligent. Turing Tests have become the holy grail of the Artificial Intelligence community and many scientists consider the challenge to be as insurmountable as superluminal space travel or nuclear fusion. . . [read more AI beats human intelligence on Yahoo answers social networking site ]
More information:
Buddhabots.com
Posted by ljmacphee on November 21, 2007 under artificial intelligence in the news, bots |
Anyone running a website has been plagued by link spam. It shows up in your access-log files, in false comments on a blog, even in user registrations on a blog. An incredible amount of resources are being put into both sides of this battle. Do a search on any search engine and you’ll find many sites in the top ten results who are just pages of links, or partially scraped content from several other sites. These sites hold little value to the users and hurt legitimate sites.
Identifying and preventing spam was cited as one of the top challenges in web search engines in a 2002 paper. Amit Singhal, principal scientist of Google Inc. estimated that the search engine spam industry had a revenue potential of $4.5 billion in year 2004 if they had been able to completely fool all search engines on all commercially viable queries. Due to the large and ever increasing financial gains resulting from high search engine ratings, it is no wonder that a significant amount of human and machine resources are devoted to artificially inflating the rankings of certain web pages. . . . [ read more Spam Rank - Full automatic link spam detection work in progress ( pdf )]
Link spam is different than regular links in that it shows up in access-logs, links in are often only found in comments on blogs, links may be hidden ( text same color as background ) or cloaked ( show users some thing different than you show search engine bots ) and often all appear in a very short period of time. Other tells include lots of links from low page rank sites or lots of links from sites with the same page rank. Page rank follows a power law and incoming links should do the same.
More information:
Transductive link spam detection (pdf)
Spam Rank - Full automatic link spam detection work in progress ( pdf )
Detecting link spam using temporal information (pdf )