Module 5: Multi Objective Optimization & Hybrid Systems

► Multi-objective optimization problem.

► Principles of Multi-objective optimization,
► Dominance and Pareto-optimality.
► Optimality conditions.
► Neuro-fuzzy hybrid systems.
► Genetic – neuro hybrid systems.
Module 5: Multi Objective Optimization & Hybrid Systems

► MOOP-Linear &Nonlinear, Convex & Non-Convex

► Principles of MOO-Illustrating Pareto Optimal Solutions
► Objectives in MOO
► Dominance & Pareto-Optimality
► Concept of Domination
► Properties of Dominance Relation, Pareto Optimality
► Procedure for finding a non-dominated set
► Optimality Conditions
► Neuro Fuzzy hybrid system-Classification& Characteristics
► Genetic –neuro hybrid systems
Multi-objective optimization problem.
5.1 Multi-objective optimization problem

► Most real-world problems involve the simultaneous optimization of

several objective functions.
► Generally, these objective functions are measured in different units, often competing
and conflicting.
► MOOP deals with more than one objective function that is to be minimized or maximized
► The MOO or multi-objective optimization refers to finding the optimal solution values of
more than one desired goal.
► These optimal solutions are known as Pareto-optimal solutions.
Multi-objective optimization problem

► MOOP Involves more than one objective function that is to be minimized or maximized
► Multi-objective optimization having such conflicting objective functions gives rise to a
set of optimal solutions, instead of one optimal solution, because no solution can be
considered to be better than any other with respect to all objectives.
► Minimizing cost while maximizingcomfort while buying a car, and maximizing
performance at the same time as minimizingfuel consumptionand emissionof
pollutants of a vehicle are examples of multi-objective optimization problems involving
two and three objectives, respectively.
Multi-objective optimization problem
MOOP Mathematical form
► A solution x is a vector of n decision
► Multi-objective optimization problems with a
variables x={x1,x2,x3,…xn}
number of objectives and a number of equality
► Associated with the problem are J
and inequality constraints can be formulated as: inequality and K equality constraints.
► The term gj(x) and hk(x) are called
constraint functions.
► If any solution x satisfies all constraints
and variable bounds, it is known as a
feasible solution. The set of all feasible
solutions is called the feasible region, or
S
Multi-objective optimization problem
formal specification of MOOP
Illustration: Decision space and objective space
► For each solution x in the
decision variable space, there
exists a point in the objective
space(Z) denoted by
f(x)=z={z1,z2,…..,zm}

► The mapping takes place

between an n-dimensional
decision vector(x) and M
dimensional objective vector(y).
Illustration: Decision space and objective space
Illustration: Single vs. multiple objectives
MOOP: Trade-off and conflicts in solutions
MOOP: Illustration: ideal solution vs. real solution
5.1.1 Linear and Non linear MOOP

► If all the objective functions and constraints are linear in the MOOP, then the problem is
defined as a Multi-Objective Linear Problem(MOLP)
► If any of the objective functions or constraints are nonlinear , then the resulting problem
is called a nonlinear Multi-Objective optimization Problem(nonlinear MOOP)
► Like linear programming problems, MOLP has many theoretical properties
► For nonlinear problems, the solution techniques often do not have convergence proofs.
► Since most of the real-world Multi-Objective optimization Problems are nonlinear in
nature, we do not assume any particular structure of the objective and constraint
functions here.
5.1.1 Linear and Non linear MOOP

► If all the objective functions and constraints are linear in the MOOP, then the problem is
defined as a Multi-Objective Linear Problem(MOLP)
► If any of the objective functions or constraints are nonlinear , then the resulting problem
is called a nonlinear Multi-Objective optimization Problem(nonlinear MOOP)
5.1.1 Linear and Non linear MOOP

► An optimization problem is nonlinear if the objective function f(x) or any of

the inequality constraints ci(x) ≤ 0, i = 1, 2, …, m, or equality constraints dj(x) = 0, j = 1,
2, …, n, are nonlinear functions of the vector of variables x.
► For example, if x contains the components x1 and x2, then the function
3 + 2x1 − 7x2 is linear, whereas the functions (x1)3 + 2x2 and
3x1 + 2x1x2 + x2 are nonlinear.
► Nonlinear problems arise when the objective or constraints cannot be expressed as
linear functions without sacrificing some essential nonlinear feature of the real-world
system
5.1.2 Convex and Nonconvex MOOP
5.1.2 Convex and Nonconvex MOOP
5.1.2 Convex and Nonconvex MOOP [another definition]

Geometrically, a function is convex if a line

segment drawn from any point (x, f(x)) to
another point (y, f(y)) -- called the chord from
x to y -- lies on or above the graph of f, as in
the picture
5.1.2 Convex and Nonconvex MOOP
5.1.2 Convex and Nonconvex MOOP

A non-convex function "curves up and down" -- it

is neither convex nor concave.
A familiar example is the sine function.
If a function g(x) is nonconvex , the set of solutions
satisfying g(x)>=0
Principles of Multi-objective optimization,
Principles of Multi-objective optimization,
Principles of Multi-objective optimization
Principles of Multi-objective optimization
Principles of Multi-objective optimization
• point A represents a solution that incurs a near-minimum cost but is highly accident-
prone (predisposed).
• On the other hand, point B represents a solution that is costly but is near the least
accident-prone.
• One cannot really say whether solution A is better than solution B or vice versa because
one solution is better than the other in one objective but is worse in the other.
• solution C is not optimal because there exists another solution D in the search space,
which is better than solution C in bothobjectives
• One cannot conclude about an absolute hierarchy of solutions A, B, D, or any other
solution in the set.
Principles of Multi-objective optimization
• These solutions are known as Pareto-Optimal solutions (named after Italian economist
Vilfredo Pareto (1906))
• The set of the best compromise solutions is referred to as the Pareto-ideal set,
characterized by the fact that starting from a solution within the set, one objective can
only be improvedat the expense of at least one otherobjective.
• In front of the Pareto-Optimal front are un-attainable solutions corresponding to the
optimal of both objectives.
• Thearea behind the Pareto-Optimal frontis known as feasible search space or feasible
design space.
Objectives of Multi-objective optimization

► A multi-objective optimization algorithm must achieve:

1. GuidethesearchtowardstheglobalPareto-Optimalfront.
2. Maintain solution diversityin the Pareto-Optimal front.
► The first goal is mandatory in any optimization task.
► When solutions converges close to the true optimal solutions, that one can be
assured of their near optimality properties.
► The second goal is entirely specific to MOOP.
► since MOEA deals with two spaces-decision variable space and objective
space-’diversity’ among solutions can be defined in both of these spaces.
Objectives of Multi-objective optimization

► For example, two solutions' are diverse, in the decision variable space, if their
Euclidean distance(length of a line segment between the two points) in the
decision variable space is large.
► Similarly, two solutions are diverse in the objective space, if their Euclidean
distance in the objective space is large.
► Diversity in one space usually means diversity in the other space, this may not
be so, in all problems.
► In such complex and nonlinear problems, it is then the task to find a set of
solutions having a good diversity in the desired space.
Dominance and Pareto-optimality
 Dominance and Pareto-optimality

► Most MOOP uses the concept of dominance in their search.

► Concept of Dominance
► Properties of Dominance Relation
► Pareto Optimality
Dominance
Dominance

◼ In a multi-objective optimization problem, the goodness of a solution is

determined by the dominance
Example of Dominance
Properties of dominance relation
Properties of dominance relation
Pareto optimal Solution
Pareto optimality

• Consider solutions 3 and 5

• Solution 5 is better than solution 3 with respect to
f1 while 5 is worse than 3 with respect to f2.
• Hence, we can not conclude that 5 dominates 3 nor
3 dominated 5.
• In other words, we can not say that two solutions 3
and 5 are better.
• {3,5} can be considered a non-dominated front
Non-dominated set
Non-dominated set
Graphical Depiction of Pareto Optimal Solution
Graphical Depiction of Pareto Optimal Solution
Pareto optimal set
Pareto optimal set
Pareto optimal set
Pareto Optimality
► Solutions along the line are all non-
dominated solutions.
► Dominated solutions are inside the line
as there is another solution on the line
with at least one objective that is better.
► The line is the Pareto-optimal front and
the solutions on it are called Pareto-
optimal.
► All Pareto-optimal solutions are
non-dominated
Example of Pareto Optimal solution
► There are 9 types of air tickets with different time and cost details.
► We need to choose from them with objective functions of minimum
cost and minimum time.
► If we compare air tickets A and B – we see that while
A is better from a Time point of view, and B is better from the Cost
angle. A and B thus form a non-dominated set.
► However, if we compare B and C, we see that B is equal to C in time,
but better than C in Cost. Hence, we can say that B “dominates” C .
► So, as long as B is a feasible option, there is no reason to choose C.
Example of Pareto Optimal solution
► There are 9 types of air tickets with different time and cost details.
► We need to choose from them with objective functions of minimum
cost and minimum time.
► Find the dominance of A, by comparing A, with B, C, D, E, F, G, H, I.
► A→B- Non dominating,
► A→C- A Dominates C
► A→D- Non dominating ,A→E- Non dominating
► A→F- Non dominating, A→G- Non dominating
► A→H- Non dominating, A→I- Non dominating
Example of Pareto Optimal solution
► Similarly Find the dominance Set of B, by comparing B, with C, D, E,
F, G, H, I.
► B→C- B Dominates C
► B→D- Non dominating- B Dominates E
► B→F- Non-dominating, B→G- B Dominates G
► B→H- B Dominates H, B→I- B Dominates I
► B dominates C, E, G, H, I
► Similarly No dominance set for C, E, G, H, I
► D Dominates E, G, H, I
► F dominates G, H, I
Example of Pareto Optimal solution

► 9 air tickets to choose from them with objective functions of minimum

cost and minimum time.

• Thus, A, B, D, and F form a “Non-dominated set”.

• Hence,A, B, D, and F will form the Pareto-Optimal Front or Non-
Dominated Front and they are the Pareto-Optimal solutions.
Example of Pareto Optimal solution
Pareto Vs Non Pareto optimization Techniques
Pareto Vs Non Pareto optimization Techniques
Pareto Vs Non Pareto optimization Techniques
Non-dominated Sorting Genetic Algorithm-I
Non-dominated Sorting
Non-dominated Sorting
Non-dominated Rank Comparison
Optimality conditions
Optimality Conditions
Optimality Conditions
Optimality Conditions
Hybrid systems
Hybrid systems
► A Hybrid system is an intelligent system that is framed by combining at least two intelligent
technologies like Fuzzy Logic, Neural networks, Genetic algorithms, reinforcement learning, etc.
► The combination of different techniques in one computational model makes these systems
possess an extended range of capabilities.
► These systems are capable of reasoning and learning in an uncertain and imprecise
environment. These systems can provide human-like expertise like domain knowledge,
adaptation in noisy environments, etc.
► Typesof Hybrid Systems:
► Neuro-Fuzzy Hybrid systems
► Neuro Genetic Hybrid systems
► Fuzzy Genetic Hybrid systems
Genetic neuro hybrid systems
Genetic neuro hybrid systems
► A Neuro Genetic hybrid system is a system that combines Neural networks: which are capable to
learn various tasks from examples, classify objects and establish relations between them, and a
Genetic algorithm: which serves important search and optimization techniques.
► Genetic algorithms can be used to improve the performance of Neural Networks and they can be
used to decide the connection weights of the inputs.
► These algorithms can also be used for topology selection and training networks.
► Applications:
► Face recognition
► DNA matching
► Animal and human research
► Behavioural system
Genetic neuro hybrid systems
Genetic neuro hybrid systems
► Working Flow:
► GA repeatedly modifies a population of individual solutions. GA uses three main types of rules at
each step to create the next generation from the current population:
► Selection to select the individuals, called parents, that contribute to the population of the next generation
► Crossover to combine two parents to form children for the next generation
► Mutation to apply random changes to individual parents in order to form children
► GA then sends the new child generation to the ANN model as a new input parameter.
► Finally, calculating the fitness by the developed ANN model is performed.
Genetic neuro hybrid systems
► Advantages:
► GA is used for topology optimization i.e. to select a number of hidden layers, number of hidden nodes, and
interconnection pattern for ANN.
► In Gas, the learning of ANN is formulated as a weight optimization problem, usually using the inverse mean
squared error as a fitness measure.
► Control parameters such as learning rate, momentum rate, tolerance level, etc are also optimized using GA.

► It can mimic the human decision-making process.

► Disadvantages:
► Highly complex system.

► Accuracy of the system is dependent on the initial population.

► Maintenance costs are very high.

Neuro-fuzzy hybrid systems
Neuro-fuzzy hybrid systems

► The Neuro-fuzzy system is based on a fuzzy system which is trained on the basis of the working
of neural network theory.
► The learning process operates only on the local information and causes only local changes in
the underlying fuzzy system.
► A neuro-fuzzy system can be seen as a 3-layer feedforward neural network.
► The first layer represents input variables, the middle (hidden) layer represents fuzzy rules and
the third layer represents output variables.
► Fuzzy sets are encoded as connection weights within the layers of the network, which provides
functionality in processing and training the model.
Neuro-fuzzy hybrid systems
Neuro-fuzzy hybrid systems

► Working flow:
► In the input layer, each neuron transmits external crisp signals directly to the next layer.
► Each fuzzification neuron receives a crisp input and determines the degree to which the input
belongs to the input fuzzy set.
► The fuzzy rule layer receives neurons that represent fuzzy sets.
► An output neuron combines all inputs using fuzzy operation UNION.
► Each defuzzification neuron represents the single output of the neuro-fuzzy system.
Neuro-fuzzy hybrid systems

► Advantages:
► It can handle numeric, linguistic, logic, etc kind of information.
► It can manage imprecise, partial, vague, or imperfect information.
► It can resolve conflicts through collaboration and aggregation.
► It has self-learning, self-organizing and self-tuning capabilities.
► It can mimic the human decision-making process.
► Disadvantages:
► Hard to develop a model from a fuzzy system
► Problems in finding suitable membership values for fuzzy systems
► Neural networks cannot be used if training data is not available.
Neuro-fuzzy hybrid systems

► Applications:
► Student Modelling

► Medical systems

► Traffic control systems

► Forecasting and predictions