15
Research Experiences for Undergraduates: Integrated Machine Learning Systems www-symbiotic.cs.ou.edu/reu Sponsored by: NSF and Oklahoma EPSCoR

Research Experiences for Undergraduates: Integrated Machine Learning Systems Sponsored by: NSF and Oklahoma EPSCoR

Embed Size (px)

Citation preview

Page 1: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Research Experiences for Undergraduates: Integrated Machine Learning Systems

www-symbiotic.cs.ou.edu/reu

Sponsored by: NSF and Oklahoma EPSCoR

Page 2: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Machine Learning

Given some data set, construct a model that can be used to interpret or to react to new data

Different machine learning problem classes:• Supervised learning: teacher knows the correct answer• Semi-supervised learning:

– Teacher can evaluate an answer, but does not necessarily know the correct answer

– Or: teacher only knows some of the answers (or some of the data)

• Unsupervised learning: there is no teacher

The boundaries are sometimes fuzzy. We often see mixtures of methods.

Page 3: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Bringing Machine Learning to the Real World

• What is the research question?

• What is the nature of your data and your learning problem?

• What are the appropriate machine learning approaches?

• How can we be sure that it works?

Page 4: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

REU Learning Goals

• Understand a range of machine learning problems and methods– Everyone will delve into more detail on some

aspects of these

• Experimental design– Experimental hypotheses– Designing an experiment to test a hypothesis– Evaluating the results of the experiment

Page 5: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

REU Learning Goals II

• Statistical tools• Professional development

– Reading, critiquing, and writing papers– Giving talks– Asking questions– Attending conferences– Graduate school– Career choices– Engineering ethics (2 full days)

Page 6: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Undergraduate Participants

Returning Students:• Josh Southerland (OU)• David Gagne (OU)

New Students:• Andy Spencer (Rose Hulman)• Rachel Shadoan (OU)• Samuel Bleckley (OU)• Benjamin Dunham (Carroll)• Peter Golbus (Bard)• Tony Liu (UNM)• Hunger Glanz (Cal Poly SLO)• Derek Tingle (Swarthmore)

Page 7: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Faculty Mentors

• Andrew H. Fagg (OU)• Dean Hougen (OU)• Amy McGovern (OU)• Rafael Fierro (UNM)• Terran Lane (UNM)

Assessment:• Theresa Cullen (OU)

Page 8: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Student Responsibilities

• Formally join a project • Attend research meetings

– Two weekly REU meetings: one technical and one professional development

– Project-specific: determined with mentor(s)

• Reading– Books & scientific papers– Some will be assigned – others you will need

to track down yourself

Page 9: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Student Responsibilities• Writing

– Project reports on the wiki (incremental goals)– REU highlights to NSF– Conference or journal submission

• Presentations– Informal status reports: every few weeks

(follows wiki project report schedule)– REU Symposium talk (~30 minutes)– Talk to students at home institution

Page 10: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

What is a scientist?

Page 11: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

What to Expect

• Research is:– an exploratory process– not like taking classes

• Participate in the research of others and ask them to participate in yours

• Do some reading every day• Your research path is (in part) your own

responsibility• Focus (most of the time)

Page 12: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Administrivia

• Summer meetings– Tuesdays and Thursdays 1:30-3:00 CDT

• Student jobs– ???: videoconference master (+ need a

backup)– ???: OU social coordinator

Page 13: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Technical Meeting Topics

• The research process• AI/ML/Robotics• Probability and statistics• Reinforcement learning• Supervised learning and function

approximation• Bayesian Networks• Clustering• Evolutionary Computation

Page 14: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR

Next Meetings

• Thursday: machine learning taxonomy

• Next Tuesday: the “Art” of Research

Page 15: Research Experiences for Undergraduates: Integrated Machine Learning Systems  Sponsored by: NSF and Oklahoma EPSCoR