Introduction to Soft Computing
Hard Computing
• Coined by Prof. L.A.Zadeh, University of California, USA, in
1996
Steps to be Followed to Solve Engineering Problem:
• Variables are identified and classified into two groups – input/
condition variables (antecedents) and output/action variables
(consequents)
• Input – Output relationships are expressed using mathematical
equations (say differential equation)
• Differential equations are solved analytically or using
numerical methods
• Control action is decided based on the obtained solutions
• Hard computing is nothing but the steps stated above
Examples :
1. Stress Analysis using FEM
2. Determination of gain values of PID controller
Features of Hard Computing
• Based on pure mathematics
• Yields precise solutions
• Suitable for problems which are easy to model mathematically
• May not be suitable to solve complex real-world problems
Soft Computing
• Introduced by Prof. Zadeh, in 1992.
• Family consisting of some biologically-inspired techniques,
such as Fuzzy Logic (FL), Neural Network (NN), Genetic
Algorithm (GA) and their various combined forms, namely GA-
FL, GA-NN, NN-FL, GA-FL-NN; in which precision is traded
for tractability, robustness, ease of implementation and a low
cost solution.
GA
GA-FL GA-NN
GA-
FL-
NN NN
FL
NN-FL
Features of Soft Computing
• Does not require an extensive mathematical formulation of the
problem
• May not be able to yield so much precision as obtained by hard
computing
• Functions of the constituent members are complementary in
nature
• Control algorithms developed based on Soft Computing may be
robust and adaptive in nature
Examples:
•FL- or NN- based motion planners for intelligent and
autonomous robots
Hybrid Computing
•Combination of the conventional hard computing and
emerging soft computing
Hard Hybrid Soft
Computing Computing Computing
• A part of a complex real-world problem will be solved
using hard computing and the remaining part can be tackled
utilizing soft computing
• Here, hard computing and soft computing are
complementary to each other
Examples:
• Optimal design of machine elements using FEM and Soft
Computing
• PID controller trained by soft computing
Note:
No fighting among hard computing and soft computing people
Introduction to Intelligence
Concept of Intelligence was initiated since ancient times (Greek
Civilization, Egyptian Civilization) – Aristotle – Rene Descartes
First Project on AI: Bayesian Logic in 1956
US DARPA project: to develop intelligent Robots
Intelligence is defined as one property with the help of which, a person can
take the decision in an adaptive way, as the situation demands
Artificial Intelligence vs. Human Intelligence
Artificial Intelligence is incorporated into the Head of an Agent through
training/learning carried out by Human Intelligence, so that it (Agent) can
perform its assigned task more efficiently using its in-built sensor(s)
Example: Sniffer Dog vs. Police Man; Director vs. Actor in Film-making
Four types of AI: Reactive Machines; Limited Memory; Theory of Mind;
AI with self-awareness (dealing with consciousness)
Fields of applications of AI: Healthcare; Business; Education; Finance;
Law; Manufacturing (Automation, Machine Vision); Banking;
Transportation; Natural Language Processing; Speech Recognition;
Sentiment Analysis; Drug Design; Human Facial Recognition
Advantages of AI: Reduced time for data-driven tasks, consistent results
Disadvantages of AI: Expensive (computationally), Requires deep technical
expertise, Limited supply of qualified workers to build AI tools, Tries to
imitate the results of brain processing instead of probing the mechanisms
that give rise to the results
Computational Intelligenec (CI)
Computational Intelligence (CI): AI using Soft Computing
Soft Computing (Zadeh, 1992): Does not require extensive
mathematical formulation of the problems, uses combined
nature-inspired tools to solve practical problems in efficient
way
Biological Adaptation: Evolution and Learning
Evolution Tools: Biologically-inspired optimization tools like
GA, GP, ES, EP, DE, CA, PSO, ACO, BA, BO etc.
Learning Tools: Neural Networks; Fuzzy Reasoning Tools
Soft Computing: Combined tools like Genetic-Fuzzy System;
Genetic-Neural System; Neuro-Fuzzy System; Genetic-Neuro-
Fuzzy System
