CMSI 4320 - Cognitive Systems Design

Professor Andrew Forney • Andrew.Forney[at]lmu.edu • Spring 2024


Welcome to CMSI-4320 1 / 9 / 24

If you're reading this then you've successfully found the course page for CMSI 4320! Check this page frequently for announcements that are relevant to the course, including notes, homework assignments, and practice problems.

First things first, please read through the course syllabus located here (or in the Materials tab above).


Lecture Notes
Date Lecture Subject
Lecture 12-2
3 / 30 / 24
Counterfractionals
(That's not a real term, but suits the topic:) Beginnings of probabilistic, practical, and empirical counterfactuals... with practice!

  Lecture Video (13-2)
  Lecture Video (13-1)
Lecture 12-1
4 / 2 / 24
Imaginationland
Our first steps intuiting, and then formalizing, the final layer of the causal hierarchy: counterfactuals!

  Lecture Video (12-2)
Lecture 11-1
3 / 26 / 24
The Awesome Power of the Coin Flip
Feel like experimenting? Observational vs. experimental data and their situation on the causal hierarchy.

  Lecture Video (12-1)
Lecture 9-1
3 / 14 / 24
Adjust Cause
I ain't afraid of no ghosts (in a dataset). Unobserved confounding, identifiability, and adjustment.

  Lecture Video (11-1)
  Lecture Video (10-2)
Lecture 8-2
3 / 12 / 24
Feeling Functionless?
How to still use SCMs when you don't know the structural equations! Causal Bayesian Networks herein.

  Lecture Video (10-1)
Lecture 8-1
3 / 5 / 24
A Cause for Celebration
...because today we ascend beyond animalistic associations and dip our toes into causal modeling!

  Lecture Video (9-2)
  Lecture Video (9-1)
  Lecture Video (8-2)
Lecture 7-2
2 / 22 / 24
Climbing the Causal Ladder
A review of the associational wrung with Bayesian Networks, plus motivations for the next steps!

  Lecture Video (8-1)
  Lecture Video (7-2)
Lecture 6-1
2 / 13 / 24
RL in RL
Some of the bleeding-edge in RL, including deep-Q-learning, inverse-RL, and others!

  Lecture Video (7-1)
  Lecture Video (6-2)
Lecture 5-2
2 / 8 / 24
Potpou-RL-i
A few best practices in RL, including: Sparse vs. Dense Rewards, Optimistic Sampling, and Policy Search!

  Lecture Video (6-1)
Lecture 5-1
2 / 6 / 24
And Now, Our Feature Presentation...
Feature-based Representations and Approximate Q-Learning!

  Lecture Video (5-2)
  Lecture Video (5-1)
Lecture 4-1
1 / 30 / 24
More Q's than A's
Reinforcement learning heats up! Partially-specified MDPs, exponential moving averages, temporal difference learning, and q-learning.

  Lecture Video (4-2)
Lecture 3-2
1 / 25 / 24
The Value of Time
Value-iteration and the solutions it brings to those janky expectimax trees!

  Lecture Video (4-1)
  Lecture Video (3-2)
Lecture 2-2
1 / 19 / 24
Great Expectations
Formalizing traits of MDPs, including discounting, expectimax search, and more!

  Lecture Video (3-1)
Lecture 2-1
1 / 16 / 24
Markov's Back
Tell a friend... Motivation for Markov Decision Processes.

  Lecture Video (2-2)
Lecture 1-2
1 / 11 / 24
Us Animals Had to Start Somewhere
Motivations for Reinforcement Learning, a little gambling, and snapshots of what's to come!

  Lecture Video (2-1-2)*
  Lecture Video (2-1-1)*
* Online class while I was sick, split into 2 parts because of S-tier internet.
Lecture 1-1
1 / 9 / 24
Cogito, ergo... cogito
A preview into the course's topics, with motivating examples to whet our appetites.

  Lecture Video (1-2)
  Lecture Video (1-1)

Assignments
Number Points Date Assigned Date Due
Final Project: Jam Pac'd Final Project Score 2 / 12 / 24 4 / 30 / 24 - Code Freeze
5 / 2 / 24 - Presentations & Competition
(See spec for other deadlines)
Assignment 4 - Back Doors, Do, and other Funny Sounding Things 100 Points 3 / 28 / 24 4 / 18 / 24
Assignment 3 - Deep Pac-ets 100 Points 2 / 18 / 24 3 / 21 / 24
Assignment 2 - The Best Policy: Not Always Honesty 100 Points 1 / 30 / 24 2 / 22 / 24
Assignment 1 - MAB About You 100 Points 1 / 17 / 24 2 / 1 / 24