Question

Computational Intelligence or Soft Computing, is a new advanced information processing technique that uses characteristics closely associated with human intelligence. Some common algorithms are Genetic Algorithms, Neural networks etc. Make a report on some of these algorithms and explain their advantages and usage areas. (2 paragraphs each algorithm at least)

Answer 1

Solution :-

Various algorithms are used on Computational Intelligence or Soft Computing. They are -

1) Neural Network -

A network of artificial neurons which are inspired from biological network of neurons and as information processing units uses mathematical models in order to discover patterns in data is called as Neural Network. These patterns in data are too complex that cannot be noticed by human. Neural networks can learn and model non-linear and complex relationships because many of relationships between inputs and outputs are complex as well as non-linear. No restrictions are being imposed by Neural networks on input variables.

Neural networks are used in different areas. They are used in Image processing and character recognition. It is also used in forecasting which is extensively required in everyday business decisions, in finance and stock market, in monetary and economic policy. Neural networks are also used in data processing such as clustering, filtering, compression and blind signal separation. Regression analysis or functional approximation that includes time series prediction and modeling are some of the fields that fall under neural networks.

2) Fuzzy Logic -

The vagueness of a solution can be understood by a technique called as fuzzy logic. This solution is presented with a degree of vagueness which is practical to human secision. It is flexible and simple. With imprecise and incomplete data problems can be handled. Non linear functions having arbitrary complexity can be modeled using fuzzy logic. Fuzzy logic are customizable in natural language terms and covers a wide range of operating conditions. Also fuzzy logic is cheaper to develop.

Fuzzy Logic is applicable in various fields.Fuzzy logic can be used in control brakes in hazardous situations which depend upon speed of car, its acceleration, speed of wheel etc. It is used in auto transmission where fuel injection and ignition based on throttle setting, RPM, cooling water temperature can be controlled. It can also be used in controlling elevators where waiting time is reduced based on passenger traffic. Also fitness of the employees can be checked by using fuzzy rules.

3) Genetic algorithms (GA)-

The heuristic search and optimization techniques where the process of evolution is mimic is called as genetic algorithm. For both machine learning applications and optimization genetic algorithms are efficient and effective techniques. Multi-objective optimization is supported by GA. They are easily parallelised. It doesn't search from a single point rather it searches from a population of points. GAs are good for noisy environments.

GA is applicable to wider range of learning and optimization problem in many domains. Problems like optimization problems, skill based employee allocation problems, job shop problem, scheduling of job shop problems etc can be solved by using GA. Multidirectional search can be performed by GA by maintaining a population of potential problems.

4) Evolutionary Computation -

This algorithm is inspired by biological evolution and belongs to a family of optimization algorithms such as Genetic algorithm, Ant Colony Optimization, Particle Swarm Intelligence, Artificial Bee Colony Optimization etc. An initial set of candidate solutions is generated and updated iteratively in evolutionary computation. With respect to noisy evaluation functions this algorithms are robust. Any aspect of this algorithm can be customized and changed.

Evolutionary Computation can be used in construction of fuzzy systems, neuro fuzzy systems and neural networks. They are inspired by biological evolution such as reproduction, recombination etc.