Home > Meeting Topics and Material, News Stories > Weekly QUEST Discussion Topics and News, 21 Feb

Weekly QUEST Discussion Topics and News, 21 Feb

Weekly QuEST Discussion Topics and News

21 Feb 2014

Been a very interesting QuEST week – topics that have consumed my bandwidth include those below – I will be prepared to discuss any of them or other items of interest to those attending or phoning in:

I gave the ‘What is QuEST’ lecture to the Cognitive Modeling brown bag this week. Very interesting Ron/Sandy/I can give you what we heard from them. Walk-aways include there are obvious places for collaboration but probably should investigate specific projects/programs to do them under.

Sandy asked some questions about Case Based Reasoning – including sending me the quote: ‘Ludwig Wittgenstein, prominent philosopher whose voluminous manuscripts were published posthumously, observed that natural concepts, such as tables and chairs are in fact polymorphic and cannot be classified by a single set of necessary and sufficient features but instead can be defined by a set of instances (i.e. cases) that have family resemblances [Watson, 1999, Wittgenstein, 2010]. ‘ So a discussion on this view and its relationship to situations/qualia might be fruitful.

I gave a talk for Engineers week for AFIT/WSU/UD and the local IEEE group: ‘From Idea to invention to productization – Capt Amerika discusses experiences in the fight against Breast Cancer’: This is an open forum discussion on the journey of becoming passionate about a problem, developing an idea on how to help with the problem, inventing a new approach using that idea, garnering the resources necessary to mature the idea and then making a product and commercializing the solution to impact the maximal number of people. The experiences discussed are associated with the fight

against breast cancer specifically how a professor of electrical engineering and some of his students / colleagues with NO business experience at all successfully made this journey resulting in a public company and products that result in earlier detection of

breast cancer. – so I have this material approved for release and happy to discuss with any QuEST people if they have questions

Jared found an interesting article on ‘Black Swans’ – the misuse of statistical approaches – it was interesting and might generate some useful discussion: I came across this blog article from the height of the financial crisis in 2008: http://edge.org/conversation/the-fourth-quadrant-a-map-of-the-limits-of-statistics It gives a pretty convincing (mathematical) argument that mirrors our thoughts on unexpected queries based on Morley’s four-quadrant assessment… Basically: if you are trying to make a decision in an environment with complex payoffs where there might be “black swans,” DO NOT rely on statistical methods — they will fail miserably at some point and notions like standard deviation and variance are nearly meaningless. I particularly like the turkey example — every day for the first 1,000 day of its life, a turkey might accumulate evidence that humans care about its welfare and it might estimate very high confidence in the validity of its models — but the 1,001st day does not go so well.

I had a great running discussion via email with a group on ‘Theory of Knowledge’ culminating in a whiteboard discussion where we generated

some interesting ideas on what such a theory might provide us – Andres is the keeper of the notes from that discussion but the discussion included: Theory of Knowledge – What would it look like? Given attributes of a given inference task (what is going on = perception, what happened before = recollection, what is going to happen next = projection) estimate the impact of the human (or set of humans), the computer decision aide (or set of computer decision aids) and the mixing function that accounts for redundancy is performance as well as detractions associated with fusing the two pieces. – Example: Breast cancer detection – given attributes of the problem space (textures / displays of x-rays / performance of existing

human visual recognition tasks and computer learning approaches for similar machine vision tasks) estimate what human performance should be for ‘h’ and for ‘c’ and for ‘m’, then via taking some small amounts of data confirm your hypothesis versus doing a complete Bayesian clinical trial with bounds of probability estimating performance.- Example 2: given a new sensor (LIDAR) estimate relative dominance in h versus c versus m for the resulting capability -Note: M that is a function of h, c and the inference task is dominated by the situational representation mismatch between the inference task situational representation and the situational representation of the h and the c respectively

Mike R asked questions last week on QuEST and big data – so I pulled material from our previous discussions on big data – it is below :

WeeklyQUESTDiscussionTopicsandNews21Feb

WeeklyQUESTDiscussionTopicsandNews21Feb

Advertisements
  1. No comments yet.
  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: