You are on page 1of 9

Unit 1

Introduction to Soft Computing


Topics
• Computing
• Hard computing features
• Soft computing
• Characteristics of Soft Computing
• Examples of soft computing
• Hard computing vs soft computing
• Hybrid computing
• Major areas of Soft Computing
Computing

Y=f(x)

Input Output

• Input mapped to output using the function f(x) which is the algorithm
• Process of mapping inputs to outputs is called computation
Features:
• Precise solution
• Prime focus on precision and certainty
• Involves exact mathematical models
• Mapping form input to output unambiguous
• Can become inefficient when solving complex real world problems
• Based on conventional algorithms which require exact input data and methodology to provide
precise output
• Can fail if input data not exact
• Can be time consuming often

Termed as Hard computing by L.A Zadeh in 1996


Some examples of Hard computing
• Solving roots of polynomials
• Sorting and searching techniques
• Computational geometry (finding the shortest path in a graph)
What is Soft Computing?
Term coined by L.A. Zadeh
Collection of methodologies that aim to exploit the tolerance for imprecision
and uncertainty to achieve robustness and low cost solutions
Features
• Does not require any exact mathematical model
• Deals with approximate models
• Gives solutions to complex real problems using approximate models
• Tolerant to imprecision, uncertainty
• Tolerant to and partial truth and approximations
• Uses some biologically inspired methodologies
• Models cognitive behavior of human mind; think logically, learn from
mistakes
Examples of soft computing
• Handwritten character recognition (solved efficiently using ANN)
• Money Allocation problem(Genetic algorithm or Probability reasoning)
• Robotic movement (fuzzy logic)
Hard computing vs Soft computing
Hard Computing Soft Computing
Requires precisely stated analytical model and often Tolerant of imprecision uncertainty, partial truth and
involves a lot of computation time approximation
Based on binary logic, numerical analysis, crisp Based on fuzzy logic, neural nets, and probability
system and software reasoning
Has the characteristic of precision Has the characteristics of approximation
Requires exact input data Can deal with ambiguous and noisy data
Deterministic Probabilistic
Produces precise answers Yields approximate results
Hybrid computing
• combination of hard computing and soft computing
• Some part of the problem using hard computing
• Some part where we can tolerate an approximate solution is solved using the
soft computing

Major areas of Soft Computing


Defined as a collection of techniques spanning many different fields
• Artificial Neural Networks
• Fuzzy Systems
• Genetic Algorithm
• Probability Reasoning

You might also like