Multi-Objective

Optimization
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

2
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
xi(L)  xi  xi(U ) , i =1, 2,L, n
lower upper
bound bound

3
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

4
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

5
 Example Dominance Test
f2
(minimize) 2

4
1
5

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 Pareto-
optimal front

7
 Graphical Depiction of
Pareto Optimal Solution
f2(x)
x2 (minimize)
feasible
objective
A space
feasible
decision
B
space Pareto-optimal front
C

x1
Pareto-optimal solutions f1(x)
(minimize)
8
Goals in MOO
◼ Find set of solutions as close as possible to Pareto-
optimal front
◼ To find a set of solutions as diverse as possible

f2(x)
feasible
objective
1
space

2
Pareto-optimal front

f1(x) 9
Classic MOO
Methods
Weighted Sum Method
◼ Scalarize a set of objectives into a single objective
by adding each objective pre-multiplied by a user-
supplied weight

F( x) = m=1 wm f m ( x),
M
minimize
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
 Itcannot find certain Pareto-optimal solutions in
the case of a nonconvex objective space

12
Weighted Sum Method
(Convex Case)

f2
Feasible
objective
space
Pareto-
optimal
w1 front
w2
f1
13
Weighted Sum Method
(Non-Convex Case)

f2
Feasible
objective
space
Pareto-
optimal
front

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
Feasible
a objective
b space

f1
1a 1b 16
-Constraint Method
◼ Advantage
 Applicable to either convex or non-convex
problems

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

17
Weighted Metric Method
◼ Combine multiple objectives using the
weighted distance metric of any solution from
the ideal solution z*

minimize lp ( x) =  m=1 wm f m ( x) − z m* 
p 1/ p
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

18
Weighted Metric Method
p=1 p=2
f2 a f2 a

z* b z* b

f1 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 Elitist MOEAs

◼ Vector evaluated GA ◼ Elitist Non-dominated

◼ Vector optimized ES Sorting GA
◼ Weight based GA ◼ Distance-based Pareto
◼ Random weighted GA GA
◼ Multiple-objective GA ◼ Strength Pareto GA
◼ Non-dominated Sorting ◼ Pareto-archived ES
GA
◼ Niched Pareto GA
25
Identifying the Non-Dominated Set
◼ Critical Step in Elitist Strategies
◼ Kung’s et. al. Method
 Computational the most efficient method known
 Recursive

26
Kung’s 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 non-
dominated 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 Kung’s Method

f2 a b c d
(min) e f g
h

f1 (min)

30
 Example of Kung’s Method
 recursively call the function ‘front’
Step 1 Step 2  front returns M as output
a T a, b, T a,b T aa a
a
b a, b, e, B bb b
e, c
e c, f, h B e,c T ee e
c a, e e B cc
d, g c
f B f, h, T T ff f
h f, h f, h B hh
d, g h
d a, e, h B T dd d
g f, h d, g d, g B gg 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
 Classifythe solutions into a number of mutually
exclusive equivalent non-dominated sets

(min) f2
F = ∪ 
j =1
Fj

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

(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
 Setnew population Pt+1 =  and set i = 1
 Until |Pt+1| + |Fi| < N, perform Pt+1 = Pt+1  Fi
and increase i

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

Non-dominated F1
sorting
Pt+1
Pt F2

F3
discard
Qt Crowding
distance
sorting
Rt

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 non-
dominated set, some Pareto-optimal solutions may give
their places to other non-Pareto-optimal solutions
N=7

A Pareto-optimal solution is discarded 41