This course provides the student with an understanding of the basic tools of connectionist modeling. Topics covered include Hopfield nets, perceptrons, and recurrent layered networks, together with supervised and unsupervised training algorithms for such networks. We also examine some applications within psychology and cognitive science generally. Students should be able to program, and should be familiar with basic calculus and linear algebra. There will be two simulation-based homework sets, and a final exam.

Here is an introductory handout for the course.

Class notes:
January 4 lecture: Introduction and some history.
January 8 lecture: Learning in perceptrons.
January 11 lecture: Back-propagation.
January 16 lecture: Demo, no notes. But please bring the back-propagation homework assignment.
January 18 lecture: Language learning: syntax. (No notes, but here are references).
January 23 lecture: Some applications of back-propagation.
January 25 lecture: Some problems with back-propagation.
January 30 lecture: Learning sequences.
February 1 lecture: Localist and distributed representations.
February 5 lecture: Connectionist modeling and cognitive theory: the case of visual word recognition (no notes).
February 7: Discussion (no notes).
February 13 lecture: Structured connectionism (no notes).
February 15 lecture: NOTE: 2 sets of notes this time: BP wrap-up and Hopfield networks as associative memories.
February 20 lecture: Hopfield nets and energy.
February 22 lecture: Interactive activation models (no notes)
February 27 lecture: Unsupervised learning.
March 1 lecture: Reinforcement learning.
March 6: Review session.
March 8: ** In-class final exam. **

Some C programs that may be helpful.

When	Where	Who	What
Tue. Jan. 9 12 noon	BSLC 205	Terry Regier	The Emergence of Words
Mon. Feb. 19 4:30pm	Beecher 102	Morten Christiansen	Language Evolution as the Adaptation of Linguistic Structure
Tue. Feb. 20 12 noon	BSLC 205	Morten Christiansen	Individual Differences in Sentence Processing: The Importance of Experience