Pattern recognition as a foundational CT skill
Posted On Thursday, April 8, 2010 at at 4/08/2010 02:37:00 PM by Dave Burrell
Brook Hall submits the following questions:
Dear Colleagues, In anticipation of this week's busy agenda, I feel it inappropriate to ask my questions about my learning curve during the meeting. However, I have attached a small summary of an exercise I am composing for my Zoology class this Fall. It is not the exercise itself for which I need some feedback, but the questions it has brought to mind.
I think the most important is "What do you foresee a module containing?" I am looking for a 'skeleton' to mold my thoughts. How long? What are the components? What will success look like, etc...Lesson plans are a snap. How do they fit in a module? Or do they?
My energies are high and my creativity robust. I am simply looking for a tiny road map to follow and need some metaphorical "climbing tools" for my mountain.
Looking forward to hearing the reports on Friday
Dan Ross answers:
Humans are great at pattern recognition, but computers are less great at it. We CAN get computers to do pattern recognition and there are a variety of techniques to do this, such as nueral networks and other kinds of algorithms. There is a lot of work going on by specialized computer scientists in this area. So, developing a pattern recognition tool may be too advanced for the non-specialist. However, USING some pattern recognition tools may be a more universal “foundational skill”. So, I think that the CT part of this may be:
1) Getting students to realize that computers can do pattern recognition.2) Getting students familiar with the various pattern recognition techniques that are available, and their capabilities, characteristics, and limitations.3) Getting students to select some existing PR technique and map/adapt its associated algorithm templates to their problem/data set.
Subject: Various
CS Concept: Expert Systems/Artificial Intelligence Techniques
Grade Level: Various
Lecture Unit: Instructor will explain the basic idea of rule-based expert systems using sample rules from some familiar human experts such as gardeners or medical doctors.
Assignment: Students will create a list of rules of the form if(Boolean expression) then (action), that characterize a set of knowledge rules for the body of knowledge being studied in that particular subject. For example: “If the grass is brown, then water it” for horticulture, or “If it hurts when I laugh, then don’t laugh" for pre-med students. Students will then role play and hand execute the system.
Evaluation: Role play in small groups and hand execute the list and see if it correctly diagnoses the medical condition. Send the student a bill after their claim is denied by the insurance company.
Phil Tierney answers:
An obvious biology analog would be a dichotomous key.
K-5
Tree, bush, or herb?
If tree, does it have needles, leaves or something else?
If it has needles, do they come off in bundles?
Are the bundles in 3s or 5s
Etc.
HS/13-14 level studentsBuild out a key for:
one of the simpler herbaceous genera
or a ubiquitous and familiar family (Rosaceae)
periodic table?
