Mathematical model for surprise !

[Royston]

Please take a look at the link below...

 

http://www.eurekalert.org/pub_releases/2005-11/uosc-scs112805.php

 

"At the foundation of our model is a simple theory which describes a principled approach to computing surprise in data streams. While surprise is not a new concept it had lacked a formal definition, broad enough to capture the intuitive meaning of the term, yet quantitative and computable…. Beyond vision, computable surprise could guide the development of data mining, as it can in principle be applied to any type of data, including visual, auditory or text."

 

I've read the article a couple of times, and although it doesn't go into considerable depth I'm still baffled by how something as subjective (or apparently not) as suprise can be modelled mathematically...if anyone would like to reiterate or elaborate on this article, I'd be very interested.

[ashennell]

Many textbooks actually describe Shannon's information as equivalent to suprise given that it is inversely proportional to the probability of an event occurring. However, this measure of suprise is not context-dependant (subjective). I think that the conditional probability has been used to obtain context-dependant measures of information but it would seem logical to make use of Bayes theorem. This seems to be what these people have done.

 

I've had a quick look what they have done seems to make sense. They are using the Bayesian framework for reasoning with uncertain information. In the Bayesian framework the prior probability distribution represents our expectations 'prior' to observing an event. They are measuring suprise as the change in this distribution following the observation of an event.i.e the extent to which an observation changes our expectations.

 

Thanks for posting this , it's something I should definately read. Once I've had a proper read through maybe I will be able ot make complete sense of it.

[Royston]

Thanks for the the explanation Ashennell, it's provided a bit more clarity. However I'm going to have to read up on the methods used to fully understand how they arrived at their conclusions.

[Royston]

Thanks for the the explanation Ashennell, it's provided a lot more clarity.

[insane_alien]

3x + 2y= ahh! simple! hehe