IT/CS%20811%20Principles%20of

1
IT/CS 811 Principles of Machine Learning and
Inference
2. Projects and assignments
Prof. Gheorghe Tecuci
Learning Agents Laboratory Computer Science
Department George Mason University
2
Individual or group projects
Journal paper
Research report
Overview of a special machine learningand/or
inference topic
Development of a learning and reasoning agent
Other
3
Example project Journal paper

• Cristina Boicu
• Exception-driven knowledge base refinement
• Prerequisites
• significant part of the research already done
  (e.g. system development)
• potential of theoretical and/or experimental
  results.
• Approach
• start with selecting the journal
• develop preliminary table of context during the
  first two weeks
• schedule section writing during each week
• - have paper ready for review two weeks prior to
  project deadline
• - paper ready for submission by the project
  deadline.

4
Example project Research report (journal
potential)

• An overview of interactive learning methods
• Prerequisites
• desire to perform an extensive literature
  search
• interest in detailed reading and comparison of
  several papers.
• Approach
• start with selecting an appropriate journal
• perform literature search
• develop comparison criteria and preliminary
  table of context
• schedule paper reading and summarization during
  each week
• - have paper ready for review two weeks prior to
  project deadline.

5
Ex. project Development of a learning and
reasoning agent

• An assistant for selecting a PhD advisor
• (could be a single person or a group project)
• Approach
• start from the assistant developed in
  IT803/Spring 2002
• develop a consistent object ontology
• develop elicitation scripts for students,
  advisors, PhD coordinator
• train the agent
• write a report/paper.
• (to be detailed based on the number of student
  participants)

6
Sample assignments
Review a project journal paper or reportand
present the review in class
Literature search on the applications of a
specific learning method (e.g. decision tree
learning) and class presentation
Develop an interesting exercise based on
somelearning algorithm (suitable for the closed
book partof the final exam) and class
presentation
Read an interesting paper on learning and
inference and present it to the class
Others

• Approach
• each assignment has a different number of points
  (based on the work required)
• accumulate 100 points in assignments.

7
Grading policy discussion

• There will be several assignments, a final exam
  and an optional project.
• The final exam will contain two parts
• a closed book one containing theoretical
  questions,
• an open book one containing problems.
• The open book part of the exam may be substituted
  by a project.
• The final grade will be computed as follows
• Class participation and assignments 20
• Closed book part of the final exam 30
• Open book part of the final exam or the project
  50