Lecture 9:

Multi-Objective
Optimization

Suggested reading: K. Deb, Multi-Objective Optimization using Evolutionary

Algorithms, John Wiley & Sons, Inc., 2001

Multi-Objective Optimization
Problems (MOOP)
Involve more than one objective function that
are to be minimized or maximized
Answer is set of solutions that define the best
tradeoff between competing objectives

General Form of MOOP

Mathematically
min/max f m ( x),

m = 1, 2,L, M

subject to g j ( x) 0, j = 1, 2,L, J
hk ( x) = 0, k = 1, 2,L, K
( L)
loweri

bound

xi x

(U )
upperi

bound

, i = 1, 2,L, n

Dominance
In the single-objective optimization problem,
the superiority of a solution over other
solutions is easily determined by comparing
their objective function values
In multi-objective optimization problem, the
goodness of a solution is determined by the
dominance

Definition of Dominance

Dominance Test

x1 dominates x2, if
Solution x1 is no worse than x2 in all
objectives
Solution x1 is strictly better than x2 in at least
one objective
x1 dominates x2
x2 is dominated by x1

Example Dominance Test

f2
(minimize)

4
1
5
3

f1 (maximize)

1 Vs 2: 1 dominates 2
1 Vs 5: 5 dominates 1
1 Vs 4: Neither solution dominates
6

Pareto Optimal Solution

Non-dominated solution set

Given a set of solutions, the non-dominated
solution set is a set of all the solutions that are not
dominated by any member of the solution set
The non-dominated set of the entire feasible decision
space is called the Pareto-optimal set
The boundary defined by the set of all point mapped
from the Pareto optimal set is called the Paretooptimal front

Graphical Depiction of
Pareto Optimal Solution
f2(x)

x2

(minimize)
feasible
objective
A
feasible
decision

space

space

Pareto-optimal front
C

Pareto-optimal solutions

x1

f1(x)
(minimize)
8

Goals in MOO

Find set of solutions as close as possible to Paretooptimal front

To find a set of solutions as diverse as possible
f2(x)

feasible
objective

space

2
Pareto-optimal front

f1(x)

Classic MOO
Methods

Weighted Sum Method

Scalarize a set of objectives into a single objective

by adding each objective pre-multiplied by a usersupplied weight

minimize

F ( x ) = m =1 wm f m ( x ),
M

subject to g j ( x ) 0,

j = 1, 2, L , J

hk ( x ) = 0,

k = 1, 2, L , K

xi( L ) xi xi(U ) ,

i = 1, 2, L , n

Weight of an objective is chosen in proportion to

the relative importance of the objective

11

Weighted Sum Method

Advantage
Simple
Disadvantage
It is difficult to set the weight vectors to obtain a
Pareto-optimal solution in a desired region in the
objective space
It cannot find certain Pareto-optimal solutions in
the case of a nonconvex objective space
12

Weighted Sum Method

(Convex Case)
f2

Feasible
objective
space

Paretooptimal
front

w1
w2
f1

13

Weighted Sum Method

(Non-Convex Case)
f2

Paretooptimal
front

Feasible
objective
space

f1

14

-Constraint Method

Haimes et. al. 1971

Keep just one of the objective and restricting the rest
of the objectives within user-specific values
minimize

f ( x ),

subject to

f m ( x) m ,

m = 1, 2, L , M and m

g j ( x ) 0,

j = 1, 2, L , J

hk ( x ) = 0,

k = 1, 2, L , K

xi( L ) xi xi(U ) ,

i = 1, 2, L , n
15

-Constraint Method
Keep f2 as an objective Minimize f2(x)
Treat f1 as a constraint

f1(x) 1

f2

a
b

1a

1b

Feasible
objective
space

f1
16

-Constraint Method

Advantage
Applicable

to either convex or non-convex

problems

Disadvantage
vector has to be chosen carefully so that it
is within the minimum or maximum values of the
individual objective function

The

17

Weighted Metric Method

Combine multiple objectives using the

weighted distance metric of any solution from
the ideal solution z*
1/ p
p
*
m

minimize lp ( x ) = m =1 wm f m ( x ) z

subject to g j ( x ) 0,
j = 1, 2, L , J
M

hk ( x ) = 0,
( L)
i

k = 1, 2, L , K

xi x

(U )
i

i = 1, 2, L , n
18

Weighted Metric Method

p=2

p=1
f2

f2

a
z*

a
z*

b
f1

b
f1

(Weighted sum approach)

19

Weighted Metric Method

p=
f2

a
b
z*
f1

(Weighted Tchebycheff problem)

20

Weighted Metric Method

Advantage
Weighted

Tchebycheff metric guarantees finding all

Pareto-optimal solution with ideal solution z*

Disadvantage
Requires

knowledge of minimum and maximum

objective values
Requires z* which can be found by independently
optimizing each objective functions
For small p, not all Pareto-optimal solutions are obtained
As p increases, the problem becomes non-differentiable
21

Multi-Objective
Genetic Algorithms

Advantages of GAs over Traditional

Methods

Our desired answer: a set of solutions

Traditional optimization methods operate on a

candidate solution

Genetic algorithms fundamentally operate on a set

of candidate solutions

23

Multi-Objective EAs (MOEAs)

There are several different multi-objective
evolutionary algorithms
Depending on the usage of elitism, there are two
types of multi-objective EAs

24

Multi-Objective MOEAs
Non-Elitist MOEAs

Vector evaluated GA
Vector optimized ES
Weight based GA
Random weighted GA
Multiple-objective GA
Non-dominated Sorting
GA
Niched Pareto GA

Elitist MOEAs

Elitist Non-dominated
Sorting GA
Distance-based Pareto
GA
Strength Pareto GA
Pareto-archived ES

25

Identifying the Non-Dominated Set

Critical Step in Elitist Strategies
Kungs et. al. Method

Computational

the most efficient method known

Recursive

26

Kungs et. al. Method: Step 1

Step 1: Sort population in descending order of
importance of the first objective function and
name population as P
Step 2: Call recursive function Front(P)

27

Front(P)
IF |P| = 1,
Return P
ELSE
T = Front ( P(1: [ |P|/2 ]) )
B = Front ( P( [ |P|/2 + 1 ] : |P|) )
IF the i-th non-dominated solution of
B is not dominated by any nondominated solution of T, M=T {i}
Return M
END
END
28

Notes on Front (P)

|| is the number of the elements
P( a : b ) means all the elements of P from
index a to b,
[] is an operator gives the nearest smaller
integer value

29

Example of Kungs Method

f2
(min)

h
f1 (min)
30

Example of Kungs Method

n recursively call the function front

Step 1
a
b
e
c
f
h
d
g

Step 2

o front returns M as output

a, b, e,
c, f, h
d, g
a, e, h

T a, b,
e, c
a, e

T a,b

B f, h,
d, g

f, h

T
B
T

e,c

B
T

f, h f, h

B
T

d, g d, g

aa
bb
ee
cc
ff
hh
dd
gg

a
b
e
c
f
h
d
g
31

Elitist MOEAs

Elite-preserving operator carries elites of a

population to the next generation
Rudolph(1996) proved GAs converge to the
global optimal solution of some functions in
the presence of elitism
Elitist MOEAs two methods are often used

Elitist Non-dominated Sorting GA (NSGA II)

Strength Pareto EA

* Reference:

G. Rudolph, Convergence of evolutionary algorithms in general search spaces, In

Proceedings of the Third IEEE conference of Evolutionary Computation, 1996, p.50-54

32

Elitist Non-Dominated Sorting GA

(Deb et al., 2000)

Use an explicit diversity-preserving strategy together

with an elite-preservation strategy

33

Elitist Non-Dominated Sorting GA

Non-Dominated Sorting
Classify

the solutions into a number of mutually

exclusive equivalent non-dominated sets

(min) f2

F =

F
j
j =1

F3
F2

F1

(min) f1

34

Elitist Non-Dominated Sorting GA

Determine Crowding Distance

Denotes

half of the perimeter of the enclosing

cuboid with the nearest neighboring solutions in
the same front
Cuboid
(min) f2
i

(min) f1

35

Elitist Non-Dominated Sorting GA

Crowding tournament selection

Assume that every solution has a non-domination

rank and a local crowding distance
A solution i wins a tournament with another
solution j
1. if the solution i has a better rank
2. They have the same rank but solution i has
a better crowing distance than solution j

36

Elitist Non-Dominated Sorting GA

Step 1
Combine

parent Pt and offspring Qt populations

Rt = Pt Qt
Perform a non-dominated sorting to Rt and
find different fronts Fi

Step 2
new population Pt+1 = and set i = 1
Until |Pt+1| + |Fi| < N, perform Pt+1 = Pt+1 Fi
and increase i
Set

37

Elitist Non-dominated Sorting GA

Step 3
Include

the most widely spread solutions (N-|Pt+1|)

of Fi in Pt+1 using the crowding distance values

Step 4
Create

offspring population Qt+1 from Pt+1 by using

the crowded tournament selection, crossover and
mutation operators

38

Elitist Non-Dominated Sorting GA

Pt

Non-dominated
sorting

F1
F2

F3

Qt
Rt

Pt+1
discard
Crowding
distance
sorting

39

Elitist Non-dominated Sorting GA

Advantages

The diversity among non-dominated solutions is

maintained using the crowding procedure: No
extra diversity control is needed
Elitism protects an already found Pareto-optimal
solution from being deleted

40

Elitist Non-dominated Sorting GA

Disadvantage
When

there are more than N members in the first nondominated set, some Pareto-optimal solutions may give
their places to other non-Pareto-optimal solutions
N=7

A Pareto-optimal solution is discarded

41

Strength Pareto EA (SPEA)

Zitzler & Thiele., 1998

Use external population P

Store fixed number of non-dominated solutions

Newly found non-dominated solutions are
compared with the existing external population and
the resulting non-dominated solutions are
preserved

42

SPEA Clustering Algorithm

1.
2.
3.

Initially, each solution belongs to a distinct cluster Ci

If number of clusters is less than or equal to N, go to 5
For each pair of clusters, calculate the cluster distance
dij and find the pair with minimum cluster-distance
d12 =

4.
5.

1
C1 C2

d (i, j )

iC1 , jC 2

Merge the two clusters and go to 2

Choose only one solution from each cluster and remove
the other (The solution having minimum average
distance from other solutions in the cluster can be
chosen)
43

SPEA Clustering Algorithm

Cluster 1

N=3

f2
Cluster 2

Cluster 3

f1
44

SPEA Algorithm

Step 1. Create initial population P0 of size N

randomly and an empty external population P0 with
maximum capacity of N
Step 2. Find the non-dominated solutions of Pt and
copy (add) these to Pt
Step 3. Find the non-dominated solutions of Pt and
delete all dominated solutions
Step 4. If |Pt| > N then use the clustering technique
to reduce the size to N
45

Step 5 Fitness evaluation

Elites: assign fitness to each elite solution i by

using
Si =

# of population members dominated by elite i

N +1

For current population: assign fitness to a current

population member j
F j = 1 + Si
iD j

where Dj is the set of all external population

members dominating j
Note: a solution with smaller fitness is better
46

SPEA Algorithm Fitness Eval.

f2

1
9/7
a

14/7

N= 6, N=3
2/7

9/7
2/7 7/7

3
2
3/7

10/7

10/7

f1
# of current population members dominated by a = 2
Sa =
7
N +1

F1 = 1 + Si
iD j

2 9
= 1+ =
7 7
47

SPEA Algorithm

Step 6
Apply

a binary tournament selection (in a

minimization sense), a crossover and a mutation
operator to Pt U Pt and generate the next population
Pt+1

48

Advantages of SPEA
Once a solution Pareto-optimal front is found,
it is stored in the external population
Clustering ensures a spread among the nondominated solutions
The proposed clustering algorithm is
parameterless

49

Disadvantages of SPEA

A balance between the regular population size

N and the external population size N is
important
If

N is too large, selection pressure for the elites is

too large and the algorithm may not converge to
the Pareto-optimal front
If N is too small, the effect of elitism will be lost

50