CS 561: Artificial Intelligence
Instructor:
Sofus A. Macskassy, [email protected]
TAs:
Nadeesha Ranashinghe (
[email protected]
)
William Yeoh (
[email protected]
)
Harris Chiu (
[email protected]
)
Lectures:
MW 5:00-6:20pm, OHE 122 / DEN
Office hours:
By appointment
Class page:
http://www-rcf.usc.edu/~macskass/CS561-Spring2010/
This class will use
http://www.uscden.net/
and class webpage
- Up to date information
- Lecture notes
- Relevant dates, links, etc.
Course material:
[AIMA] Artificial Intelligence: A Modern Approach,
by Stuart Russell and Peter Norvig. (2nd ed)

This
** preview**
has intentionally

**sections.**

*blurred***to view the full version.**

*Sign up*
CS561 - Lecture 21 - Macskassy - Spring 2010
2
Temporal Probability Models [Ch
15]
Time and uncertainty
Inference: filtering, prediction, smoothing
Hidden Markov models
Kalman filters (a brief mention)
Dynamic Bayesian networks
Particle filtering

CS561 - Lecture 21 - Macskassy - Spring 2010
3
Time and uncertainty
The world changes; we need to track and predict it
Diabetes management vs vehicle diagnosis
Basic idea: copy state and evidence variables for each time
step
X
t
= set of unobservable state variables at time
t
e.g.,
BloodSugar
t
,
StomachContents
t
, etc.
E
t
= set of observable evidence variables at time
t
e.g.,
MeasuredBloodSugar
t
,
PulseRate
t
,
FoodEaten
t
This assumes
discrete time
; step size depends on problem
Notation:
X
a:b
=
X
a
,
X
a+1
, … ,
X
b-1
,
X
b

This
** preview**
has intentionally

**sections.**

*blurred***to view the full version.**

*Sign up*
CS561 - Lecture 21 - Macskassy - Spring 2010
4
Markov processes (Markov chains)
Construct a Bayes net from these variables: parents?
Markov assumption
:
X
t
depends on
bounded
subset of
X
0:t-1
First-order Markov process
:
P
(
X
t
j
X
0:t-1
) =
P
(
X
t
j
X
t-1
)
Second-order Markov process
:
P
(
X
t
j
X
0:t-1
) =
P
(
X
t
j
X
t-2
,
X
t-1
)
First-order:
Second-order:
Sensor Markov assumption
:
P
(
E
t
j
X
0:t
,
E
0:t-1
) =
P
(
E
t
j
X
t
)
Stationary
process: transition model
P
(
X
t
j
X
t-1
)
and
sensor model
P
(
E
t
j
X
t
)
fixed for all
t