Instructor: Michael Coen (mhcoen@cs.wisc.edu)

Lectures: Tuesdays and Thursdays, 11AM-12:15PM

Location: tba

This introductory class addresses two basic questions:

1) How can understanding intelligence in animals lead to the

development of more sophisticated computational systems?

2) How can biologically-inspired machine learning lead to new theories

of learning and cognition in animals?

We'll examine how these two questions are surprisingly interrelated,

using formal methods from cognitive science, artificial intelligence,

machine learning, computational biology, and cognitive neuroscience.

Along the way, we'll examine applications from a variety of domains,

including artificial life, concept and language acquisition,

programmatic trading on Wall Street, and automatically segmenting

brain regions based on function.

Prerequisites: Because this class is interdisciplinary, students from

all backgrounds are welcome. Some mathematical background would be

helpful, particularly linear algebra and calculus, but the course will

be essentially self-contained. Also helpful would be previous

exposure to basic concepts in computer science and programming in a

high-level language, such as Matlab or Python. (E.g., CS302 would be

more than sufficient.) If you have any questions about prereqs,

please contact me; we'll make every effort to accommodate students

from a variety of disciplines.

Readings: Papers and chapters will be drawn from the literature in

cognitive science, machine learning, artificial intelligence,

cognitive neuroscience, computational biology, and philosophy.

Assignments: 4-5 problem sets and a final project. No exams.

-----------------------------

Topics to be covered include:

-----------------------------

1) Cognition in the Wild

* Learning outside of computers

* Why animal and machine intelligences rarely overlap

* Historic ramifications of confusing the two

* Examining cognitive capabilities of animals

* The empiricist school of psychology

* Evolutionary biology vs. mathematical notions of simplicity

2) Recasting machine learning in a biological context

* Occam's Razor

* Bayesian learning and statistical inference

* Graphical models and learning about causality

* Learning categories

* Common sense reasoning

3) Supervised, semi-supervised, and unsupervised learning -- what

happens in the real world?

* The phylogenetic bias -- nature vs. nurture debate for cognition

* Developmental learning; Chomsky's Poverty of Stimulus framework

* Language acquisition; statistical models of language

* Concept formation

4) Decision making

* Computationally modeling how animals make decisions

* Understanding "irrational" behavior; expected value problems;

irrelevant distractors; modeling terrorists

* Applications to game theory. What is bluffing? Can computers

play poker?

5) Swarm Intelligences

* How do systems of simple components produce complex behaviors?

* Modeling insect colonies; modeling ecologies; Lotka-Volterra

equations

* Searle's Chinese Room problem

* Genetic algorithms for swarms and evolutionary dynamics

6) Cognitive neuroscience

* Biologically-inspired artificial intelligence

* Functional and regional specificity in the brain

* Introduction to fMRI, MEG, EEG, EIT, and their applications

* Segmenting brain structures using biologically-inspired machine

learning