Progress on the CRCD Experiences at the University of Central Florida An NSF Project

1
Progress on the CRCD Experiences at the
University of Central FloridaAn NSF Project

• M. Georgiopoulos (), E. Gelenbe (), R. DeMara
  (), A. Gonzalez (),
• M. Kysilka (), M. Mollaghasemi (), A. Wu (),
• G. Anagnostopoulos (), I. Russell (), J.
  Secretan ()
• University of Central Florida(), Imperial
  College ()
• Florida Institute of Technology (), University
  of Hartford()

2
Presentation Outline

• CRCD People
• CRCD Goals/Vision
• CRCD Accomplishments
• Machine Learning I and II Classes
• Machine Learning Projects
• Assessment and Evaluation Results
• Conclusions
• Future Challenges

3
CRCD People

• CRCD PI Faculty (Georgiopoulos (EE), Gelenbe
  (CS), DeMara (CpE), Gonzalez (CpE), Kysilka
  (Education), Mollaghasemi (IEMS), Wu (CS)
• CRCD Graduate Students (3 CpE, 1 EE, 1 CS, 1
  IEMS)
• CRCD Advisory Board Academia (2 CS, 5 EECE, 1
  Math)
• CRCD Advisory Board Government/Industry (Los
  Alamos (1), NASA AMES (2), SAIC (1), Soar (1)).

4
CRCD in Machine Learning _at_UCF

• Integrating Machine Learning Concepts into the
  Undergraduate Integrating Curriculum
• http//www.ece.ucf.edu/crcd
• Machine learning is concerned with building
  computer systems that have the ability to improve
  their performance in a given domain through
  experience
• The CRCD project involves a comprehensive
  approach to the development of a model for the
  integration of Machine Learning throughput the
  entire engineering and science curriculum. The
  goal is increased exposure to Machine Learning
  technology for a wider range of students in
  science and engineering than is currently
  available

5
CRCD in Machine Learning _at_UCF

• Project Goals
• Incorporate current state-of-the-art Machine
  Learning research results into the undergraduate
  and first year graduate curriculum in a way that
  enhances the students critical thinking,
  intellectual growth and communication skills
• Offer a unique curriculum development, by
  traditional undergraduate standards, where
  faculty integrate their current research results
  into the curriculum. This curriculum will be
  interesting and dynamic, reflecting changes in
  the facultys and the Machine Learning
  community's research interests over time.

6
CRCD in Machine Learning _at_UCF

• Project Goals
• Offer the opportunity to a multi-disciplinary
  group of students (spanning the entire
  Engineering spectrum of electrical, computer,
  industrial, civil, mechanical, as well as
  computer science students) to benefit from this
  innovative research and curriculum development
• Assess and evaluate the impact of our efforts
  through a sequence of carefully chosen evaluation
  instruments, developed by our education
  specialist
• Disseminate the curriculum development efforts to
  other Universities

7
CRCD Vision
8
CRCD AccomplishmentsActivities-Tools

• Developed 8 Machine Learning Modules
• We taught these modules in lower level
  undergraduate classes and graduate classes
• Created 3 Java Applets
• Introduced two senior level Machine Learning
  courses (Machine Learning I and II)
• Machine Learning I has beet taught twice by now
  (Fall 03 and Fall 04)
• Machine Learning II has been taught twice by now
  (Spring 04 and Spring 05)
• Educational material for Machine Learning I and
  II classes have been and are being produced

9
CRCD AccomplishmentsEducational Outreach

• CRCD Board 13 members (8 academia, 5
  industry/government)
• A Board meeting has taken place in April of 2005
• Significant interaction with University of
  Hartford and Florida Tech
• Co-authorship of papers
• An NSF CCLI AI proposal was funded (University
  of Hartford)
• A Prototype CCLI-EMD proposal was funded (Florida
  Tech)

10
CRCD AccomplishmentsEducational Outreach

• 10 graduate students (7 Ph.D.s, 3 Masters)
• 4 Masters students have graduated
• One Ph.D. has graduated in the Spring of 2004
• CRCD Marketing to over 350 U/G Students (modules,
  senior design classes, person-to-person
  interaction)
• More than 20 U/G students exposed to Machine
  Learning research in Machine Learning I and II

11
CRCD AccomplishmentsPeople - Ideas

• Jimmy Secretan, a CRCD undergraduate student in
  03-04
• Graduated with a B.S. degree in the Spring of
  2004
• He is currently a Ph.D. student in the CpE
  program at UCF
• He has published three conference papers, and he
  has one journal paper under review
• His CRCD work has motivated some of the work that
  he is planning for his Ph.D.
• He is currently supervising some of the 04-05
  CRCD students

12
CRCD AccomplishmentsPeople-Ideas

• Jose Castro, a CRCD graduate student in 01-03
• Has graduated in the Spring 2004
• Has published numerous conferences
• Has one journal published, and one currently
  under review
• He is currently a faculty member at the
  Technological Institute of Costa Rica and the
  Director of a Research Institute in Informatics

13
CRCD AccomplishmentsPublications -- Ideas

• Due to partial support from the CRCD project one
  of the CRCD Ph.D. students has published
• Many conferences in related to Machine Learning
  conferences
• Published one journal paper
• Has one journal paper under review (co-authored
  by a CRCD U/G student)

14
Machine Learning I Course

• Contents of ML-I
• Inductive Learning
• ART Neural Networks
• Random Neural Networks
• Experimental Design and Simulation
• Genetic Algorithms
• Evolvable Hardware

15
Machine Learning II Course

• Machine Learning II is project oriented
• Projects are assigned based on material covered
  in ML-I
• Examples of ML-II projects (04-05)
• CRCD Project on Parallel FAM
• CRCD Project on Evolving Control for Distributed
  Micro-Aerial Vehicles (MAVs)
• CRCD Project on HSFC Nearest Neighbor

16
MAVs Project
17
MAVs Project
18
HSFC Nearest Neighbor Project
19
HSFC Nearest Neighbor Project
20
HSFC Nearest Neighbor Project
21
Parallel FAM Project FAM Architecture
22
Parallel FAM Project Learning in FAM
23
Parallel FAM Project Learning in FAM
24
Parallel FAM Project Learning in FAM
25
Parallel FAM Project Covertype Database
26
Parallel FAM ProjectPipelined FAM
27
Parallel FAM ProjectSpeed-up of Pipelined FAM
28
Assessment and Evaluation of ML-I and ML-II
Question a I found the individual time I spent
with my professor helpful while I was working on
my project Question f My professors were
available to help me when I needed it as I worked
on my project
29
Assessment and Evaluation of ML-I and ML-II
Question b I found I could get help from my
fellow students while working on my project
30
Assessment and Evaluation of ML-I and ML-II
Question c The initial lectures in the ML-II
class helped me to decide on the project chosen
31
Assessment and Evaluation of ML-I and ML-II
Question d The ML-I's lectures helped me as I
worked on my project
32
Assessment and Evaluation of ML-I and ML-II
Question e I found the idea of the portfolio of
work helpful as I worked on my project
33
Assessment and Evaluation of ML-I and ML-II
Question g I like having to complete a project
as part of the class assignments Question h I
liked the way the class was structured (initially
lectures, then work on project, monthly
presentations etc.) Question i I found the idea
of providing input about the progress of my
work (on a monthly basis) helpful for me and my
other classmates
34
Assessment and Evaluation of ML-I and ML-II
Question j I would have preferred if the project
was assigned to me earlier
35
Assessment and Evaluation of ML-I and ML-II
Question k I found the papers that were assigned
for me to read helpful in completing the project
36
Assessment and Evaluation of ML-I and ML-II
Question l I would be willing to take another
class like this
37
Assessment and Evaluation of ML-I and ML-II
Question m I think all the professors were
effective in explaining their ideas
38
Successes of Recruiting Strategies

• Populated our ML-I class with approximately 20
  students (where fundamentals of our research were
  discussed)
• Populated our ML-II class with approximately 20
  students (where students worked on Machine
  Learning projects)

39
Successes of the ML classes

• Five Conference Papers have been published or
  accepted for publication
• Three conference papers are currently under
  review
• One journal paper is currently under review
• Two more journal papers are submitted for
  publication this summer (Summer 05)

40
Challenges for CRCD

• Recruiting Challenges
• Staffing Challenges
• Research Challenges

41
Thank you for attendingthis presentation
42
Questions???