Hard & soft computing

Hard and Soft Computing; Introduction to ANS

1. Hard & soft computing S. CAROLINE, Assistant Professor, ECE / SXCCE

2. Hard Computing Hard computing is a traditional computing. It requires a precisely stated analytical model and usually a lot of computation time. It strictly follows known steps to solve a task as opposed to  soft computing  which is  heuristic . Based on clearly written algorithm (structured) Based on mathematical formulae Mathematical formula ---- algorithm --- program Calculates and performs statics Stores in memory , retrieves the data, edit, monitor and control Intelligence is missing

3. Hard computing is achieved using sequential  programs  that use binary logic. It is deterministic in nature. The input data should be exact and the output will be precise and verifiable. Advantages Accurate solutions can be obtained Faster Disadvantages Not suitable for real world problems

4. Soft Computing Soft computing is the  use of approximate calculations to provide imprecise but usable solutions to complex computational problems. Soft computing represents the certitude that the human mind has the capability to store and process information that is imprecise and lacks certainty. Idea is to model the cognitive behavior of brain Brain has a better remembrance and reasoning capacity Way to mimic human intelligence and convert it to program

5. Soft computing is used for approximate models to give solution to complex problems. This is in contrast with hard computing which deals with precise models providing accurate solutions. Prof Lotfi Zadeh introduced the term, Soft Computing. The objective was to emulate human mind as closely as possible. The word, soft means flexible, adjustable, random, vague, approximate, imprecise, perceivable and non-deterministic. Fusion of soft and hard computing techniques are also useful in applications such as robotics.

6. Advantages Robustness Low cost Rapport with reality Ability to solve complex problems

7. Applications Internet search technique based on genetic algorithm Hybrid fuzzy controllers Rocket engine control Semantic web Data compression Audio recording Speech recognition Image understanding

8. Introduction to ANS Technology Architecture of the human brain is significantly different from the architecture of a conventional computer The response time of the individual neural cells is typically on the order of a few tens of milliseconds The massive parallelism and interconnectivity observed in the biological systems evidently account for the ability of the brain to perform complex pattern recognition in a few hundred milliseconds

9. In many real-world applications, we want our computers to perform complex pattern recognition problems. Since our conventional computers are obviously not suited to this type of problem, we therefore borrow features from the physiology of the brain as the basis for our new processing models The technology has come to be known as artificial neural systems (ANS) technology, or simply neural networks .

10. Storage capacity of computer is in the range of GHz , so it takes time to go through all databases in case of face recording In brain, immediate occurances are stored in superficial layer. Hence we identify everything easily.

11. Neural Network Inter connection of neurons such that neuron outputs are connected through weights to all other neurons including themselves.

12. How ANN resembles brain? Knowledge is acquired by the network through a learning process Inter neuron connection strengths known as synaptic weights are used to store the knowledge

13. Structure Neural-network structure is a collection of parallel processors connected together in the form of a directed graph, organized such that the network structure lends itself to the problem being considered

14. We can schematically represent each processing element (or unit) in the network as a node, with connections between units indicated by the arcs The direction of information flow in the network through the use of the arrowheads on the connections Size the number of processing units (called hidden units) that must be used internally, to connect them to the input and output units already defined using weighted connections, and to train the network

15. A method of computing based on the interaction of multiple connected processing elements A powerful technique to solve many real world problems Ability to learn from experience in order to improve the performance Ability to deal with incomplete information