MAS.S62 Interactive Machine Learning
Fall 2013
Instructors: William Bradley Knox, Cynthia Lynn Breazeal
TAs: Nicholas Brian DePalma, Nicholas Gillian
Meeting:
W 1-2
(E14-493)
Meeting: F 11-1
(E14-493)
:
This website is being phased out. Instead, the class will use http://iml.media.mit.edu
MAS.S62 Interactive Machine Learning (0-12-0), H-Level
Instructor: Dr. W. Bradley Knox (principal), with Prof. Cynthia
Breazeal
***If you did not receive an email on 9/5 about Piazza, etc.,
contact Brad Knox.***
Many applications of machine learning involve interactions with
humans. Humans may provide input to a learning algorithm, including
input in the form of labels, demonstrations, corrections, rankings,
or evaluations. And they could give such input while observing the
algorithm's outputs, potentially in the form of feedback,
predictions, or demonstrations. Although humans are an integral
part of the learning process, traditional machine learning systems
used in these applications are agnostic to the fact that
inputs/outputs are from/for humans. However, a growing community of
researchers at the intersection of machine learning and
human-computer interaction are making interaction with humans a
central part of developing machine learning systems. These efforts
include applying interaction design principles to machine learning
systems, using human-subject testing to evaluate machine learning
systems and inspire new methods, and changing the input and output
channels of machine learning systems to better leverage human
capabilities. This course focuses on interactive machine learning
(IML), which I define to be machine learning with a human
in the learning loop, observing the result of learning and
providing input meant to improve the learning outcome.
In this research-focused course, we will explore the motivation for interactive machine learning, create and analyze a range of simple interactive machine learning systems, and cover a breadth of specific interactive machine learning problems and approaches. Much of the course will be focused on building towards and executing a final research project that involves a machine learning algorithm with a human in the loop.
Course format: The course will consist of small hands-on projects and a final research project; guest lectures by researchers of interactive machine learning and related areas; student-written reviews and discussions on readings, some of which the class will choose; flipped classroom content and lecture when necessary; student presentations on explorations of available toolkits for machine learning, visualization, etc..
Prerequisites:
• A familiarity with machine learning. We will review many machine
learning concepts, but this course is unlikely to teach the
fundamentals of machine learning to someone who is unacquainted.
Completing a MOOC course (e.g., on Coursera or EdX) on machine
learning would provide sufficient background.
• Strong programming skills
Speakers confirmed so far:
• Nick Gillian (MIT)
• Rebecca Fiebrink (Princeton)
• Simone Stumpf (City University London)
• Krzysztof Gajos (Harvard)
• Ashish Kapoor (Microsoft Research)
• Joe Konstan (Univ. of Minnesota)
• Henry Lieberman (MIT)
Announcements
No announcements
