Research On The Cooperation Of Soft Computing And Its Application In Captcha Attack

Soft computing is an association of computing methodologies that includes fuzzy logic, neuro-computing, evolutionary computing etc. The common feature of the methodologies is able to tolerate uncertainty, and partial truth data, use linear model instead of nonlinear one and use finite instead of infinite one. When unable or difficult to obtain the optimal solution of the problem, we have to settle for the second best, and find the suboptimal solution. Based on soft computing, the models are closer to the objective substance and human thinking. Between the methodologies, it is a kind of mutual cooperation rather than mutual repulsion. Researching on the cooperation of the soft computing can make their respective advantages complementary each other to play their better performance advantage.During constructing classifier of the connotative layer, constructively neural network algorithm, Covering Algorithm which select the center samples by a random method lead to instability of the classifier's classification capacity constructed by Covering Algorithm. We combine neural network with genetic algorithm and introduce the concept of competition in the process of constructing the classifier's classification. In this paper, Good point-Set genetic algorithm was used to search optimal covering from covering sets. It can eliminate the poor cover, retain the better cover and signally reduce the number of coverage and samples rejection, so raises the classification capacity of the total population. The results of the experiments prove that our algorithm has a good robustness, validity and extensibility.Aimed to solve the problem of the classifier's classification capacity encounter larger impact when processing noisy, fuzzy and uncertain data. We combine rough set with neural network and present the concept of covering entropy. At the same time, we find a method of categorical ability. That is, under the condition of ensuring the classified ability of the algorithm, the uncertainty of the classifier is decreased by reducing the covering which has the largest covering entropy in a group of coverings. The results of the experiments prove that our algorithm has a good validity and extensibility, and has a good ability of dealing with fuzzy and uncertain data. Captcha is a program which can be used to tell computers and humans apart, the image verification code is a typical application of Captcha. The development of the Captcha is summarized and the feature and design ideas of the common Captcha are also compared, then we present the attack method on common Captcha and design the system of feature extraction of Captcha.Covering algorithm and support vector machine are two important classification of machine learning. But there have not a general-purpose classifier based on covering algorithm for a long period of time. It hindered the promotion of covering algorithm to some extant. Based on the covering algorithm, a classifier, Jcover, was designed and applied in this paper. The soft computing algorithm is embedded in Jcover. The results of the experiments prove that the model achieve excellent robust and high success recognition rate.