Hypothetical Decision Tree
This, approximately, is the decision tree I use to decide if I should accept Linked In connection requests;-)
CSV file is here
A Method for More Intelligent Touch Event Processing: summary page
Problem I attempt to mitigate:
With current mobile OSs, a touch event missing a widget's bounding rectangle
by just one pixel frequently activates an unintended widget.
Below is a 'teaser' screenshot from the above summary page.
The bottom half of the screen has a semi-transparent colored pixel overlay
which indicates to which widget a touch (tap) event would be mapped if my proposed
(first-draft) algorithm were used at the OS level to map touch events to widgets.
Hint: In addition to widget bounding rectangle, my simple algorithm also considers
the distance from the touch point to the centroids of widgets.
Problem: I suspect many medical research studies do not collect data which include the causal variables thereby 'confounding' the ability of researchers to uncover important causal links leading to medical outcomes (good or bad). This project describes a graph-based platform which would support causal analysis from arbitrarily diverse information sources along with strong annotation of supposed information describing its provenance and estimated validity. The reason that the word 'unconstrained' is used is because of the inadvertent bias implicit in the budgets and minds of researchers when selecting the variables on which to collect data.
An unconstrained environmental factors health informatics platform. The diagrams below are 'cleansed.'
The diagram below is an excerpt from an RDF lifestyle/medical/occupational/environmental graph exploring how one
might go about finding patterns leading to an understanding of the causation and/or prevention of specific diseases.
In addition to providing world-class analytics algorithms, it would provide reusable information security and
researcher-population survey sub-systems so that researchers would not have to spend research dollars
on commonly required IT infrastructure.
A graph based topical and relational approach to knowledge representation and learning.
The proposed platform could have a major impact on:
Three illustrative slides from presentation:
Argument Diagram with expansion to supportive facts and provenance:
Underlying 4-tuple graph data and metadata model + Vocabulary architecture:
Common underlying technologies for both of above:
I developed a core Java in-memory RDF 'triple store' - actually a QuadStore (subject, predicate, object, uuid).
This QuadStore supports a handful of graph query operations and could easily be instrumented to work in a parallel, distributed environment via a TCP agent layer.
Other miscellaneous postings below... see my blog for additional tech-related thoughts interspersed through my posts...
Informal Observations on a Google+ Posting
A mini-presentation on an informal experiment related to a Google+ posting's statistics and, especially, the posting of a spreadsheet with statistics on the photo posting.
Here is a link to the PDF presentation
Lens Test Pattern
Here is a home-made lens test pattern. (The image below is just a low-resolution screenshot - don't use it.)
If you buy a 2' x 3' foam board, you can tape several of these sheets to cover it with test patterns. I had
Kinkos print 10 sheets on their best color printer and taped these to the foam board. I also taped a mirror
to the center of the foam board in order to ensure perpendicularity of the optical axis of my camera lens
with the foam board.
Image Processed PowerPoint Background (for light text presentations)
Random Software-Generated Background Graphics
Copyright (c) Richard Creamer 2008 - 2014 - All Rights Reserved