IA 471: Computational Psychology 1 Course information Spring 2016

advertisement
IA 471: Computational Psychology
Spring 2016
1
Course information
Class schedule
Instructor
Office hours
Textbook
2
Section 01: M/T/Th/F 1st period (8:05–8:55 am)
Section 02: M/T/Th/F 2nd period (9:00–9:50 am)
Room A219
Alan Jern
Office: B103A
Email: jern@rose-hulman.edu
Thursdays 6th–8th periods, and by appointment
Lee & Wagenmakers, Bayesian Cognitive Modeling: A Practical Course
Overview
Computational psychologists aim to develop theories of cognition and behavior that are precise
enough that they can be implemented as computer programs. To illustrate why this is a desirable
goal, consider the following theory of how people decide whether a piece of furniture is a sofa or an
armchair: People compare the item to sofas and armchairs they have previously seen. If the item
is more similar to the sofas, they call it a sofa. If the item is more similar to the armchairs, they
call it an armchair.
This sounds like a sensible theory, but if you tried to turn it into a computer program for
recognizing furniture, you would quickly run into a host of problems. What does “similar” mean?
How many sofas or armchairs do you compare the item to? Which sofas and armchairs in memory
do you compare the item to? How does this comparison process happen? As you will learn in this
class, the process of thinking computationally forces you to be explicit about hidden assumptions
in theories like this one. Moreover, the process of developing and testing a computational model
can provide insights into how the mind works that would not have been evident otherwise.
In this course, you will learn how mathematical and computational tools drawn the fields of
statistics and machine learning can be used to develop models of human learning and reasoning.
1
3
Assessment
Component
Homeworks (7)
Project
Reading responses (14)
Paper presentation
Participation
Total
3.1
Weight
35%
35%
15%
10%
5%
100%
Homeworks (35%)
There will be seven homework assignments. These assignments will generally require you to apply
some of the methods learned in the preceding week. Assignments will consist of some combination
of math problems and coding assignments.
3.2
Project (35%)
At the end of the quarter, you will complete a group project related to computational psychology.
More details about the scope and expectations for this project will be posted on Moodle. There
is time allocated during Weeks 9 and 10 for you to work on your projects in class. Additionally,
during this time, I will meet with each group at least once for an update and to provide help and
feedback.
Each group will present their projects in class during Week 10. Each group member must
submit their own project report, written in their own words, due 5/24.
Assignment
Project proposal
Presentation
Report
Total
3.3
Weight
5%
5%
25%
35%
Reading responses (15%)
You will learn about most of the specific computational models in this class by reading published
papers that we will discuss in class. To ensure that you come to class prepared for discussion, you
must submit a reading response before class on days when papers are assigned.
For each paper, I will post one or two questions online that will help to guide your reading. You
will submit a response to these questions before class on the specified date. Your response does not
need to be longer than two paragraphs. There are 14 assigned reading responses, but your score
for this category will be based only on your 12 best responses.
3.4
Paper presentation (10%)
Students will be assigned in small groups to lead one of the paper discussions. The days with
student discussion leaders are identified by asterisks (*) next to the readings in the course schedule
2
below. On your group’s assigned day, you will be expected to summarize the methods used in
the paper, the results, and to come prepared with some discussion questions for the rest of the
class. On the class day before your assigned day, I will spend half of the class period meeting with
your group to clarify details of the paper, answer your questions, and help guide your presentation.
You are required to have already thoroughly read through your assigned paper (even if you didn’t
understand every part of it) before these meetings.
3.5
Participation (5%)
Because this course involves lots of discussion of papers and different computational models, I
expect you to be an active participant, both asking and answering questions in class.
As long as you show good attendance and make a reasonable effort to contribute to the class
when appropriate, you will receive full participation credit. I will warn you in advance by email if
I feel your behavior is deficient in either of these respects. That means that if you don’t hear from
me, you can assume you are on track to receive full credit. If you continue to make an inadequate
participation effort after a warning, you will receive a 0 for the participation component of your
grade.
3.6
Final grade
Grades will be assigned as follows.
Percentage
≥ 90%
87–89%
80–86%
77–79%
70–76%
67–69%
60–66%
< 60%
4
4.1
Grade
A
B+
B
C+
C
D+
D
F
Course policies
Late assignments
For the entire course, you will have two free late days that can be used for homework assignments ONLY. Homeworks will be considered one day late if they are submitted any time after
the start of class on the due date up to 24 hours later. Homeworks will be considered two days
late if they are submitted any time between 24 and 48 hours after the the start of class on the due
date. You don’t need to notify me in advance if you plan to use one of your late days—I will keep
track of your late days and notify you by email when you have no late days remaining.
Any assignments submitted after your late days are exhausted will not be accepted. The purpose
for this policy is to ensure that I can grade your homework and post keys in a timely fashion.
3
4.2
Academic integrity
Academic misconduct will be addressed according to the policies described in the Rose-Hulman
student handbook. Academic misconduct includes: (1) submitting work that is not your own; (2)
copying ideas, words, or graphics from any source without appropriate citation; (3) misrepresenting
your work or yourself (i.e., deliberately submitting the wrong assignment or lying to explain a late
assignment); (4) collaborating with other students when this is not permitted; and (5) submitting
the same work for credit in two courses without prior consent of both instructors. If you are unsure
whether something qualifies as academic misconduct, please check with me before engaging in the
behavior.
4
5
Course schedule
The following schedule lists topics, readings, and due dates for the quarter. Pages numbers refer
to the Lee & Wagenmakers book. This schedule is tentative. Schedule changes will be announced
in class and will be posted online. I will give you plenty of notice when such changes are made.
Week
1
2
3
4
5
6
7
8
9
10
Date
3/7
3/8
3/10
3/11
3/14
3/15
3/17
3/18
3/21
3/22
3/24
3/25
3/28
3/29
3/31
4/1
4/11
4/12
4/14
4/15
4/18
4/19
4/21
4/22
4/25
4/26
4/28
4/29
5/2
5/3
5/5
5/6
5/9
5/10
5/12
5/13
5/16
5/17
5/19
Topic
Introduction
Foundations
Bayesian inference
Bayesian inference
Generalization
WinBUGS
Generalization
Bayesian inference
Concept learning
Concept learning
Concept learning
Concept learning
Concept learning
Concept learning
Sampling assumptions
Sampling assumptions
Sampling assumptions
Sampling assumptions
Sampling assumptions
Sequential learning
Sequential learning
Sequential learning
Causal learning
Causal learning
Inference
Inference
Neural networks
Neural networks
Neural networks
Neural networks
Neural networks
Model selection
Project time
Project time
Project time
Project time
Project time
Presentations
Presentations
Reading
Due
Farrell & Lewandowsky
pp. 3–7, 11–12
pp. 37–45
Tenenbaum & Griffiths
pp. 16–32
Sanjana & Tenenbaum
pp. 49–59
Response
Nosofsky
pp. 212–223
Lake et al (*)
Kemp et al (*)
Response
HW1
Response
HW2; Response
Response
Response
HW3
Shafto et al (*)
Frank & Goodman (*)
Jern & Kemp
Gureckis & Markant (*)
Response
Response
Response
Response
Brown & Steyvers (*)
Hagmayer et al
Griffiths & Tenenbaum (*)
pp. 7–11
HW4; Response
Response
HW5
Project proposal
Rumelhart & McClelland (*)
LeCun et al
pp. 101–107, 118–124
HW6; Response
Response
HW7
5
Week
Finals
Date
5/20
5/24
Topic
Conclusion
Reading
Due
Project report
6
Download