Compact Swarm Intelligent Computing And Its Applications

Bio-inspired Swarm intelligent computing is a kind of algorithms which simulate the intelligent behavior of individuals when they cooperate with others in seeking food, migration etc. it has simple principles and is easy to implement, thus, it was accepted for many reseachers and was widely used to solve all kinds of optimization problems.Most of intelligent computing algorithms acquired the optimum by iterations of large populations. These algorithms have the special requirement that is enough hardware environment. With the rapid development of computer hardware, an ordinary computer could meet requirements of these algorithms. However, these are still some engineering domains, maybe because of some special environments with limited hardware, or because of real-time requirement in some engineering projects, or because of simply hardware environments based on fault-tolerance requirements, they couldn’t afford enough hardware to run these bio-inspired intelligent computing algorithms. Under above complicated conditions, the classical bio-inspired swarm intelligent computing algorithms can’t solve these optimization problems well. Improvements and innovations should be done on traditional bio-inspired swarm intelligent computing algorithms to adapt above application domains.This study aims in compact bio-inspired swarm intelligent computing algorithms. Based on traditional bip-inspired swarm intelligent computing algorithms, a few novel compact swarm algorithms will be proposed in this paper and some improved optimization algorithms was designed with simpler procedure、 new novel searching operator with new mathemathic idea. They could run more effectively with less dependence on hardware.The subject was organized which focused on reducing computing cost and saving memory. The main contents and innovations include:(1) A quantity representation for phernomone in Quantity Ant Colony optimization System was proposed. This representaion needs modest memory, less computing cost and less parameters compared with the traditional phernomone representation. The experimental results shows that it could speed up the convergence rate to some content.(2) A novel compact cat swarm optimization algorithm was proposed based on normal distribution model,and a new differential operator was employed to replace the original mutation operator in seeking mode. the proposed algorithm was used to solve MRI image segmentation problems successfully.(3) A perturbation vector based on Gamma probabilistic distribution model was designed to describe the population of solution sets for those problems with small size samplings. individuals under this new perturbation vector will be generated more close to the reality, it maybe provide a new idea for solving optimization problems with small size samplings. A steepest gradient method was introduced to reduce the computing cost in seeking mode. And it could solve the watermarking embedding optimization problem for audio successfully.(4) A facial expression recognition with status of eyes and lip was firstly proposed, and c CSO combination with SVM was employed to solve classification of facial expression recognition in healthcare system.The project was designed around a clue based on time complexity and space complexity. Experimental results shows that all proposed algorithms had got expected effect, and they will have good application prospects.