CS343H: Honors Artificial Intelligence</a> -- Spring 2015: Resources Page
Resources for
Honors Artificial Intelligence
(cs343H)
Weeks 0 and 1: Introduction
The slides presented in class:
PDF
Is it an Agent, or just a Program?: A Taxonomy for Autonomous Agents
(
Alternative link
).
Stan Franklin and Art Graesser
Talk on the AI 100 report in Finland
Forum for Artificial Intelligence: AI and Life in 2030
Information on
pair programming
; A
video
Related lecture slides (UC Berkeley CS188):
Introduction
Related lecture videos (UC Berkeley CS188):
Introduction to AI
Week 2: Search:
The slides presented in class:
Tuesday
;
Thursday
Sven Koenig's site on
LPA* and D* lite
Andrew Ng's
A* search notes
A student found this
A* tutorial
useful.
Online Learning of Search Heuristics
Bidirectional Search That Is Guaranteed to Meet in the Middle
Some research papers related to this week's material
AlphaGo's Nature paper
AlphaGo Zero
Related lecture slides (UC Berkeley CS188):
Uninformed Search
,
Informed Search
Related lecture videos (UC Berkeley CS188):
Agents and Search
,
A* Search and Heuristics
SBS videos:
Uninformed Search
,
A* Search and Heuristics
Handout-I
,
Solution-I
,
Handout-II
,
Solution-II
Week 3: Beyond Classical Search:
The slides presented in class:
Tuesday
;
Thursday
Continuous state space learning on Aibos:
walking
;
ball control
GA applications
From a former student:
Genetic Mona Lisa
From a former student:
robot evolution
Path search in continuous environments using
RRT's
.
Some research papers related to this week's material
A constraint-based method for solving sequential manipulation planning problems
Randomized algorithm for informative path planning with budget constraints
Related lecture slides (UC Berkeley CS188):
CSPs I
,
II
Related lecture videos (UC Berkeley CS188):
CSPs I
,
II
Handout
,
Solution
Week 4: Adversarial Search
The slides presented in class:
Tuesday
;
Thursday
;
Optimizer's curse
Some
adversarial reasoning in the 2010 Super Bowl
.
And some more from
the 2012 Super Bowl
.
The University of Alberta
GAMES
group.
The Berkeley
GamesCrafters
group.
The Stanford
General Game Playing
group.
A paper showing
PacMan is NP-hard
. A
slashdot discussion
on it.
A video about
loss aversion and risk assessment by people
Some research papers related to this week's material
Evolutionary Many-Objective Optimization Based on Adversarial Decomposition
Playing Multi-Action Adversarial Games: Online Evolutionary Planning versus Tree Search
Related lecture slides (UC Berkeley CS188):
Adversarial Search
,
Expectimax Search and Utilities
Related lecture videos (UC Berkeley CS188):
Adversarial Search
,
Uncertainty and Utilities
Starting at minute 10 of
this video
is a keynote by Mike Bowling on game playing AI, featuring their recent computer poker victory against people.
SBS videos:
Alpha-Beta
Handout
,
Solution
Week 5: Markov Decision Processes
The slides presented in class:
Context
;
Tuesday
;
Thursday
Learn a reward function:
Inverse Reinforcement Learning
Some research papers related to this week's material
Graph-based Cross Entropy method for solving multi-robot decentralized POMDPs
Dynamically Constructed (PO) MDPs for Adaptive Robot Planning
Related lecture slides (UC Berkeley CS188):
MDP I
,
II
Related lecture videos (UC Berkeley CS188):
MDP I
,
II
Handout
,
Solution
Week 6: Reinforcement Learning
The slides presented in class:
Tuesday
;
Thursday
Side-Stepping of the Triple Pendulum on a Cart
RL for POMDP
Theory of Generalization: How an infinite model can learn from a finite sample
Deep RL Bootcamp
lectures.
Some research papers related to this week's material
Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learning
Deep Reinforcement Learning for Robotic Manipulation with Asynchronous Off-Policy Updates
Related lecture slides (UC Berkeley CS188):
RL I
,
II
Related lecture videos (UC Berkeley CS188):
RL I
,
II
Handout
,
Solution
Week 7: Bayes' Nets Representation and Inference
The slides presented in class:
Tuesday-Probability
;
Tuesday-Representation
;
Thursday-Independence
;
Thursday-Inference
A
Bayes' Net tool
that you can play with.
Review of probability theory
Common probability distribution
Some research papers related to this week's material
Bayesian learning for safe high-speed navigation in unknown environments
Dynamic Bayesian Network for semantic place classification in mobile robotics
Related lecture slides (UC Berkeley CS188):
Representation
,
Probability
,
Independence
,
Inference
,
Sampling
Related lecture videos (UC Berkeley CS188):
Representation
,
Probability
,
Independence
,
Inference
,
Sampling
SBS videos:
Independence
,
Sampling
,
Sampling II
,
Elimination of One Variable
,
Variable Elimination
Handout-I
,
Solution-I
,
Handout-II
,
Solution-II
Week 8: Midterm
The slides presented in class:
Tuesday-Sampling
The Berkeley course's
past exams
(with solutions).
Week 9: (Hidden) Markov Models, Particle Filters, and VOI
The slides presented in class:
Tuesday-Context
;
Tuesday-HMM
;
Tuesday-Particle Filter
Thursday-DecisionsVPI
;
Thursday-Localization
The readings on particle filters from my Fall 2015 graduate class on autonomous robots
. Some are from the book "Probabilistic Robotics" by Thrun, Burgard, and Fox.
The full list of resources from that class
, including the following:
Slides from the book: (
ppt
).
Some
videos
from the textbook authors.
Stochastic Simulation Algorithms for Dynamic Probabilistic Networks
Markov Chain vs. HMM
Some research papers related to this week's material
Vehicle localization using particle filter
3D Audio-Visual Speaker Tracking with an Adaptive Particle Filter
A good explanation of smoothing/filering
Related lecture slides (UC Berkeley CS188):
Markov Models
,
HMM
,
Particle Filters and Applications of HMMs
Related lecture videos (UC Berkeley CS188):
Markov Models
,
Applications of HMMs
,
HMMs Filtering
Handouts:
HMM
;
HMM-solution
;
VPI
;
VPI-solution
;
Decision Networks
;
Decision Networks-solution
Week 10: Naive Bayes and Perceptrons
The slides presented in class:
Tuesday-Context
;
Tuesday
;
Thursday
Naive Bayes in Python and R
From Perceptrons to Deep Networks
An explanation of
the connection between the number of bits required to encode a hypothesis and minimum description length (MDL)
Some research papers related to this week's material
Multi-face Detection System Design based on Naive Bayes Classifier
Extreme Learning Machine for Multilayer Perceptron
Being Bayesian about Network Structure: A Bayesian Approach to Structure Discovery in Bayesian Networks
Related lecture slides (UC Berkeley CS188):
Naive Bayes
,
Perceptron
Related lecture videos (UC Berkeley CS188):
Naive Bayes
,
Perceptron
SBS videos:
Maximum Likelihood Examples
,
Laplace Smoothing
,
Perceptrons
In-class exercises
;
solutions
Week 11: Deep Learning
The slides presented in class:
policy gradient
;
optimizing NN structure for RL
.
Helpful Neural Net Notes:
Backpropagation
;
Function approximation
;
CNN for Visual Recognition
Goodfellow, Bengio, and Courville's
Deep Learning book
.
Why deep learning is suddenly changing your life
Explanation on why ReLu is better
Geoffrey Hinton talk: What is wrong with CNN
Geoffrey Hinton's recent theory to replace deep learning
The UC Irvine
Machine Learning Repository
Open source classification/regression software:
WEKA
A new effort for comparing ML algorithms:
ML comp
Some research papers related to this week's material
Deep Learning
A Neural Algorithm of Artistic Style
Generating Sequences With Recurrent Neural Networks
Designing Neural Network Architectures using Reinforcement Learning
A slide deck on
TensorFlow
Related lecture slides (UC Berkeley CS188):
Deep Learning I
,
Deep Learning II
In-class exercises:
computing gradients
;
solutions
;
NN representation
;
solutions
;
Week 12: SVMs, Kernels, and Clustering
The slides presented in class:
Tuesday-Context
;
Tuesday
;
SVM-Kernels python implementation
Clustering implementation
Comparing Python Clustering Algorithms
Deep learning vs. kernel acoustic models for speech recognition
Clustering Vs. Classification on Keyword Research
How Much Training Data is Required for Machine Learning?
SVM incremental learning, Adaptation and Optimization
Some research papers related to this week's material
Deep Learning using Linear Support Vector Machines
General Tensor Spectral Co-clustering for Higher-Order Data
Related lecture slides (UC Berkeley CS188):
Kernels and Clustering
Related lecture videos (UC Berkeley CS188):
Kernels and Clustering
Some
practice problems
and
solutions
(see especially the first problem).
Week 13: Classical Planning
The slides presented in class:
Tuesday-Context
;
Tuesday
;
Tuesday-Prodigy
;
Thursday
A USC
planning class
, including some
graphplan slides
Planning today:
27th International Conference on Automated Planning and Scheduling
.
Planning competitions
Answer Set Programming
Two papers I published on domain-independent planning heuristics during my first year of grad school. A shorter
conference paper
and a longer
journal article
.
Course: Planning, Execution, and Learning
Practice
,
Solution
Week 14: Philosophical Foundations
The slides presented in class:
Thursday
The
killer drones video
from Stuart Russell.
Why the Future Doesn't Need Us
by Bill Joy - Wired, 2000. (
pdf version
)
The Essence of Soccer: Can Robots Play Too?
Peter Stone, Michael Quinlan, and Todd Hester.
Appeared in a
book on philosophy and soccer
.
Some writings on
the singularity
: a
resesarch institute
, and
Ray Kurzweil's page
.
Jordan Pollack's
GOLEM project
: evolving physical robots.
The
Lifeboat Foundation
is dedicated to assessing and protecting against threats to humanity.
The
Asilomar conference
on recombinant DNA.
Ben Kuipers' personal stance on accepting military research funding.
3 principles for creating safer AI
Can we build AI without losing control over it?
[
Back to Department Homepage
]
Page maintained by
Peter Stone
Questions? Send me
mail