Who am I?
0 Senior Software Engineer at Google, joined in 2010.
0 Ph. D. in Computer and Information Sciences from Temple University.
0 Adjunct Professor of Computer Science at Monmouth University, 2008-09.
0 Data Structures! Algorithms! Java! Life?
0 K-nearest neighbor classification? Massively multithreaded servers?
Students independently coming up with their own projects? Sneaky.
0 Pretty radical teaching philosophy, but paid off. 4 year follow-up:
0 One student turned his course project into a publication, now a researcher.
0 K-nearest neighbor project inspired one student to go into machine learning.
0 Lunch with a student at Google on Friday; he’s in the final interview round.
0 Trying to start a school: http://www.projectpolymath.org
0 Started coding at 8!
0 First thing I did was corrupt my hard drive. Six times.
0 Fascinated by the atmosphere before that – didn’t want to go out “if
cumulonimbus clouds were outside”.
0 Published medical researcher.
[email protected]
Why so much passion?
GPA vs. Career
Became a CS
Adjunct, loved it!
First CS course
4
Started programming
3.8
First AI
publication
3.6
3.4
Started number
theory research
3.2
3
2.8
2.6
First technical job
(web developer)
Started AI research
2.4
1st
2nd
3rd
4th
5th
6th
7th
8th
Technology education was my life!
9th
10th
11th
12th
Fr.
Sph.
Jr.
Sr.
MS
Math + Peer Tutor
PhD
Why is this important?
500 STEM Majors: Age of First STEM Interest
You are here
20%
2%
3%
5%
57%
"Always"
Elementary School
Middle School
High School
College
13%
Unsure
Source: http://www.microsoft.com/en-us/news/presskits/citizenship/docs/STEMPerceptionsReport.pdf
“Technical Intuition”
“I think so, Brain, but where are
we going to get string, monopoly
money, and a flock of pigeons?”
“Are you pondering what I’m
pondering, Pinky?”
0 Technical intuition: an analogy-driven understanding of a technical
problem through its relationship to what is already known.
Pigeonhole Sort
?
Problem: Sorting money in ascending order of value.
How would you do it?
Intuition and Logic
0
0
0
0
You cannot describe what happens when you learn something, and you can’t control when it happens – it’s subconscious.
Learning is an intuitive process. Acquiring that intuition is the “click”, the “light switching on”.
Repetition is not intuition. Intuition is understanding one problem, not solving 50 by blindly applying rules.
Excessive formalism often interferes with acquiring intuition, but intuition assists in understanding formalism.
Intuition and Logic
Swiss roll
Fruit rollup.
Weinberger, K.Q., Saul, L.K. Unsupervised learning of image manifolds by semidefinite
programming. CVPR 2004, Vol. 2, pp. 988-995.
http://repository.upenn.edu/cgi/viewcontent.cgi?article=1000&context=cis_papers
0
0
0
0
You cannot describe what happens when you learn something, and you can’t control when it happens – it’s subconscious.
Learning is an intuitive process. Acquiring that intuition is the “click”, the “light switching on”.
Repetition is not intuition. Intuition is understanding one problem, not solving 50 by blindly applying rules.
Excessive formalism often interferes with acquiring intuition, but intuition assists in understanding formalism.
Thinking > Memorization
0 Real-world problem solving:
Intuitively understand the problem
Create a hypothetical solution
Implement the solution
No
Test the solution
Did it work?
Yes
Peanut butter
jelly time!
CS305 Midterm, Spring 2009.
Open book, open Internet.
1. Can an algorithm simultaneously be Θ(𝑛)
and Ω log 𝑛 ? Why or why not?
2. What Look
is the complexity
of the
function
at Wikipedia
a
lot
𝑓 𝑛 = 7𝑛4 + 2𝑛3 − 20𝑛 + 6?
3. True/False: If algorithm A is Θ(𝑛2 ) and A
runs in 4 seconds when 𝑛 = 2, A will run in 9
seconds when 𝑛 = 3.
4. What is the complexity of the following
algorithm in terms of n?
…
5. What is the advantage of this type of test?
Average grade: 84%
Creativity in the Classroom
public class SingingFrog extends Frog {
public SingingFrog(Song frogSong) { … }
@Override
public void croak() { … }
}
>
“This is a constructor…”
“This is a destructor…”
“This is a method…”
0 Hands-on creative learning is critical.
0 Autodidactic programming is almost always hands-on.
0 Make lifechanging experiences for your students:
0 Playing with Capsela toys in 4th grade sparked interest in electronics and robotics.
0 Now creating brain-computer interfaces and trying to treat Parkinson’s disease.
How do you get 8-year olds to
program, anyway?
Prior Experience vs. Grade, 30 students
100
Grade
95
90
85
80
75
0
1
2
3
4
5
Years of prior experience
6
7
0 Don’t assume they can’t do it. Kids can learn very fast and tend to be very creative.
0 Bring the concepts down to their level, present them intuitively, and connect them with the known.
0 Make it a game (and let them make games).
0 Show them role models who do it. Be one if possible.
0 As teachers, point students who are exhibiting proficiency at additional resources.
0
“College for kids” programs, interest groups, hackerspaces, competitions.
0
Companies at the forefront of the field.
0
In high school, actual college courses and faculty research.
8
Recognizing a technical mindset
0 Common traits of technical padawans:
0 Insatiable curiosity, asks “why?” a lot.
0 Desire to make, to build, to improve, to customize
(“takes things/ideas apart”). “Makers”
0 Perfectionism, tries to “optimize” everything.
0 Deep focus, sometimes to the exclusion of more
mundane schoolwork (possibly “underperformers”).
0 “Geeky”, makes references to geek jargon (e.g. “grok”)
and culture. Reads xkcd.
0 More general signs of developmental potential:
0 Overexcitabilities (Dabrowski):
OE
Manifestation
Psychomotor
“Can’t sit still”, high energy, may be misdiagnosed as ADHD.
Sensory
Deep connection to music, art, food, the outdoors, etc.
Intellectual
Highly curious, builds a large knowledge base at an early age.
Imaginational
“In the clouds”, thinks about possibilities more than actualities.
Emotional
Deep sense of justice, compassion, right/wrong.
Randall Munroe, xkcd: xkcd.com/519
0 Early learning in areas outside of the curriculum (with
or without engagement inside of it).
0 Leadership/sense of purpose at an early age.
Anakin Skywalker, technical padawan.
--Star Wars, Episode 1
Obvious: Make Learning Fun
The Problem of Retention
The loss of STEM students
4500000
4000000
3500000
3000000
2500000
2000000
1500000
1000000
500000
0
Why the attrition?
0 STEM is hard.
0 STEM students need to be very motivated to survive.
0 There are few support structures in place among peers, mentors, or faculty.
0 Certain short-term motivators are “dangerous”: they don’t sustain a strong
passion:
1.
Parental forcing: this generally doesn’t work in any field.
Big difference between forcing and inspiring: is the desire on the part of the
parent or the student?
2.
3.
Top-down educational incentives: scholarships and grants aren’t cited among
strong motivators for students.
Peer pressure: high school and college peer groups are different, have different
values. If one group values STEM and a new group does not, the desire may
collapse.
Effective Motivators
0 Surprisingly, the promise of a high salary is an effective motivator.
0 (That doesn’t mean that they survive the college major)
0 Passion and hard work were ranked the strongest factors to success.
Summary
0 Intuition is a powerful tool for understanding complex
scientific and technological phenomena.
0 Presented properly, it can transcend age and
background.
0 Analogy with the known is a useful technique for
accessing this intuition.
0 Hands-on creative work is vital to effective STEM
education.
0 It is necessary to find strong support networks for
students in order to raise STEM retention rates.
Questions?