Transcript lecture_02
ECE 8443 – Pattern Recognition
LECTURE 02: BAYESIAN DECISION THEORY
• Objectives:
Bayes Rule
Minimum Error Rate
Decision Surfaces
• Resources:
D.H.S: Chapter 2 (Part 1)
D.H.S: Chapter 2 (Part 2)
R.G.O. : Intro to PR
URL:
Audio:
Probability Decision Theory
• Bayesian decision theory is a fundamental statistical approach to the
problem of pattern classification.
• Quantify the tradeoffs between various classification decisions using
probability and the costs that accompany these decisions.
• Assume all relevant probability distributions are known (later we will learn
how to estimate these from data).
• Can we exploit prior knowledge in our fish classification problem:
Are the sequence of fish predictable? (statistics)
Is each class equally probable? (uniform priors)
What is the cost of an error? (risk, optimization)
ECE 8443: Lecture 02, Slide 1
Prior Probabilities
• State of nature is prior information
• Model as a random variable, :
= 1: the event that the next fish is a sea bass
category 1: sea bass; category 2: salmon
P(1) = probability of category 1
P(2) = probability of category 2
P(1) + P( 2) = 1
Exclusivity: 1 and 2 share no basic events
Exhaustivity: the union of all outcomes is the sample space
(either 1 or 2 must occur)
• If all incorrect classifications have an equal cost:
Decide 1 if P(1) > P(2); otherwise, decide 2
ECE 8443: Lecture 02, Slide 2
Class-Conditional Probabilities
• A decision rule with only prior information always produces the same result
and ignores measurements.
• If P(1) >> P( 2), we will be correct most of the time.
• Probability of error: P(E) = min(P(1),P( 2)).
• Given a feature, x (lightness), which is a
continuous random variable, p(x|2) is
the class-conditional probability density
function:
• p(x|1) and p(x|2) describe the difference
in lightness between populations of sea
and salmon.
ECE 8443: Lecture 02, Slide 3
Probability Functions
• A probability density function is denoted in lowercase and represents a
function of a continuous variable.
• px(x|), often abbreviated as p(x), denotes a probability density function for
the random variable X. Note that px(x|) and py(y|) can be two different
functions.
• P(x|) denotes a probability mass function, and
must obey the following constraints:
P ( x) 0
P(x) 1
x X
• Probability mass functions are typically used for discrete random variables
while densities describe continuous random variables (latter must be
integrated).
ECE 8443: Lecture 02, Slide 4
Bayes Formula
• Suppose we know both P(j) and p(x|j), and we can measure x. How does
this influence our decision?
• The joint probability of finding a pattern that is in category j and that this
pattern has a feature value of x is:
p ( j , x ) P j x p x p x
j
P j
• Rearranging terms, we arrive at Bayes formula:
P j x
p x
j
P j
px
where in the case of two categories:
2
px p x
j 1
j
P j
ECE 8443: Lecture 02, Slide 5
Posterior Probabilities
• Bayes formula:
P j x
p x
j
P j
px
can be expressed in words as:
posterior
likelihood prior
evidence
• By measuring x, we can convert the prior probability, P(j), into a posterior
probability, P(j|x).
• Evidence can be viewed as a scale factor and is often ignored in optimization
applications (e.g., speech recognition).
ECE 8443: Lecture 02, Slide 6
Posteriors Sum To 1.0
• Two-class fish sorting problem (P(1) = 2/3, P(2) = 1/3):
• For every value of x, the posteriors sum to 1.0.
• At x=14, the probability it is in category 2 is 0.08, and for category 1 is 0.92.
ECE 8443: Lecture 02, Slide 7
Bayes Decision Rule
• Decision rule:
For an observation x, decide 1 if P(1|x) > P(2|x); otherwise, decide 2
• Probability of error:
P ( 2 x )
P error | x
P ( 1 x )
x 1
x 2
• The average probability of error is given by:
P ( error ) P ( error , x ) dx P ( error | x ) p ( x ) dx
P ( error | x ) min[ P (1 x ), P ( 2 x )]
• If for every x we ensure that P ( error | x ) is as small as possible, then the
integral is as small as possible.
• Thus, Bayes decision rule minimizes P ( error | x ) .
ECE 8443: Lecture 02, Slide 8
Evidence
• The evidence, p x , is a scale factor that assures
conditional probabilities sum to 1:
P 1 x P 2 x 1
• We can eliminate the scale factor (which appears on
both sides of the equation):
x 1
iff
p x 1 P ( 1 ) p x 2 P ( 2 )
• Special cases:
p x 1 p x 2 :
x gives us no useful information.
P ( 1 ) P ( 2 ) : decision is based entirely on the likelihood p x i .
ECE 8443: Lecture 02, Slide 9
Generalization of the Two-Class Problem
• Generalization of the preceding ideas:
Use of more than one feature
(e.g., length and lightness)
Use more than two states of nature
(e.g., N-way classification)
Allowing actions other than a decision to decide on the state of nature
(e.g., rejection: refusing to take an action when alternatives are close or
confidence is low)
Introduce a loss of function which is more general than
the probability of error (e.g., errors are not equally costly)
Let us replace the scalar x by the vector, x, in a d-dimensional Euclidean
space, Rd, called the feature space.
ECE 8443: Lecture 02, Slide 10
Loss Function
• Let {1, 2,…, c} be the set of “c” categories
• Let {1, 2,…, a} be the set of “a” possible actions
• Let (i|j) be the loss incurred for taking action i
when the state of nature is j
• The posterior, P ( j x ) , can be computed from Bayes formula:
P ( j x )
p ( x | j ) P ( j )
p(x)
where the evidence is:
c
p ( x ) p ( x | j ) P ( j )
j 1
• The expected loss from taking action i is:
c
R ( i | x ) ( i | x ) P ( j | x )
j 1
ECE 8443: Lecture 02, Slide 11
Bayes Risk
• An expected loss is called a risk.
• R(i|x) is called the conditional risk.
• A general decision rule is a function (x) that tells us which action to take for
every possible observation.
• The overall risk is given by:
R R ( ( x ) | x ) p ( x ) d x
• If we choose (x) so that R(i(x)) is as small as possible for every x, the
overall risk will be minimized.
• Compute the conditional risk for every and select the action that minimizes
R(i|x). This is denoted R*, and is referred to as the Bayes risk.
• The Bayes risk is the best performance that can be achieved (for the given
data set or problem definition).
ECE 8443: Lecture 02, Slide 12
Two-Category Classification
• Let 1 correspond to 1, 2 to 2, and ij = (i|j)
• The conditional risk is given by:
R(1|x) = 11P(1|x) + 12P(2|x)
R(2|x) = 21P(1|x) + 22P(2|x)
• Our decision rule is:
choose 1 if: R(1|x) < R(2|x);
otherwise decide 2
• This results in the equivalent rule:
choose 1 if: (21- 11) P(x|1) > (12- 22) P(x|2);
otherwise decide 2
• If the loss incurred for making an error is greater than that incurred for being
correct, the factors (21- 11) and (12- 22) are positive, and the ratio of these
factors simply scales the posteriors.
ECE 8443: Lecture 02, Slide 13
Likelihood
• By employing Bayes formula, we can replace the posteriors by the prior
probabilities and conditional densities:
choose 1 if:
(21- 11) p(x|1) P(1) > (12- 22) p(x|2) P(2);
otherwise decide 2
• If 21- 11 is positive, our rule becomes:
choose 1 if :
p ( x | 1 )
p (x | 2 )
12 22 P ( 2 )
21 11 P ( 1 )
• If the loss factors are identical, and the prior probabilities are equal, this
reduces to a standard likelihood ratio:
choose 1 if :
p ( x | 1 )
p (x | 2 )
ECE 8443: Lecture 02, Slide 14
1
Minimum Error Rate
• Consider a symmetrical or zero-one loss function:
0
( i j )
1
i j
i j
i , j 1, 2 ,..., c
• The conditional risk is:
c
R ( i x ) R ( i j ) P ( j x )
j 1
c
P ( j x )
ji
1 P ( i x )
The conditional risk is the average probability of error.
• To minimize error, maximize P(i|x) — also known as maximum a posteriori
decoding (MAP).
ECE 8443: Lecture 02, Slide 15
Likelihood Ratio
• Minimum error rate classification: choose i if: P(i| x) > P(j| x) for all ji
ECE 8443: Lecture 02, Slide 16
Minimax Criterion
• Design our classifier to minimize the worst overall risk
(avoid catastrophic failures)
• Factor overall risk into contributions for each region:
R [ 11 P ( 1 ) p ( x | 1 ) 12 P ( 2 ) p ( x | 2 )] d x
R1
[ 21 P ( 1 ) p ( x | 1 ) 22 P ( 2 ) p ( x | 2 )] d x
R2
• Using a simplified notation (Van Trees, 1968):
P1 P ( 1 ); P2 P ( 2 )
I 11 p ( x | 1 )d x ; I 12 p ( x | 2 )d x
R1
R1
I 21 p ( x | 1 )d x ; I 22 p ( x | 2 )d x
R2
ECE 8443: Lecture 02, Slide 17
R2
Minimax Criterion
• We can rewrite the risk:
R P111 I 11 P2 12 I 12 P1 21 I 21 P2 22 I 22
• Note that I11=1-I21 and I22=1-I12:
R P111 (1 I 21 ) P2 12 I 12 P1 21 I 21 P2 22 (1 I 12 )
We make this substitution because we want the risk in terms of error
probabilities and priors.
• Multiply out, add and subtract P121, and rearrange:
R P1 21 P111 P111 I 21 P2 12 I 12 P1 21
P1 21 I 21 P2 22 P2 22 I 12
P1 21 P2 22 P2 ( 12 22 ) I 12
[ P111 I 21 P1 21 P1 21 I 21 P111 ]
P1 21 P2 22 P2 ( 12 22 ) I 12 P1 ( 21 11 )(1 I 21 )
ECE 8443: Lecture 02, Slide 18
Expansion of the Risk Function
• Note P1 =1- P2:
R 21 (1 P2 ) P2 22 P2 ( 12 22 ) I 12
(1 P2 )( 21 11 )(1 I 21 )
21 21 P2 P2 22 P2 ( 12 22 ) I 12
(1 P2 )( 21 11 )(1 I 21 )
21 ( 11 12 )(1 I 21 )
P2 [( 22 21 ) ( 12 22 ) I 12 ( 11 21 ) I 21
21 11 21 11 I 21 21 I 21
P2 [( 22 21 ) ( 12 22 ) I 12 ( 11 21 ) I 21
11 (1 I 21 ) 21 I 21
P2 [( 22 21 ) ( 12 22 ) I 12 ( 11 21 ) I 21 ]
ECE 8443: Lecture 02, Slide 19
Explanation of the Risk Function
• Note that the risk is linear in P2:
R 11 (1 I 21 ) 21 I 21
P2 [( 22 21 ) ( 12 22 ) I 12 ( 11 21 ) I 21 ]
• If we can find a boundary such that the second term is zero, then the minimax
risk becomes:
R mm ( P2 ) 11 (1 I 21 ) 21 I 21 11 ( 21 11 ) I 21
• For each value of the prior, there is an
associated Bayes error rate.
• Minimax: find the maximum Bayes error for
the prior P1, and then use the
corresponding decision region.
ECE 8443: Lecture 02, Slide 20
Neyman-Pearson Criterion
• Guarantee the total risk is less than some fixed constant (or cost).
• Minimize the risk subject to the constraint:
R ( i x ) d x constant
(e.g., must not misclassify more than 1% of salmon as sea bass)
• Typically must adjust boundaries numerically.
• For some distributions (e.g., Gaussian), analytical solutions do exist.
ECE 8443: Lecture 02, Slide 21
Summary
• Bayes Formula: factors a posterior into a combination of a likelihood, prior
and the evidence. Is this the only appropriate engineering model?
• Bayes Decision Rule: what is its relationship to minimum error?
• Bayes Risk: what is its relation to performance?
• Generalized Risk: what are some alternate formulations for decision criteria
based on risk? What are some applications where these formulations would
be appropriate?
ECE 8443: Lecture 02, Slide 22