A Simple Fuzzy Neural Classification Model Based On Unbalanced Data Of Hybrid Processing Scheme

Nowadays,artificial intelligence has been risen to the national strategy.Data which is regarded as an important factor to promote the intelligent development of artificial intelligence plays an important role in promoting the development of artificial intelligence.Many problems that people face in daily life can be easily transformed into data classification problems.The neuro-fuzzy classifier,which is infiltrated and cooperated with each other by neural networks and fuzzy inference systems,is a popular method for solving data classification problems.Due to the rapid development of Internet technology,the interrelationships between things and things,and the interference of too many uncertainties in real life,the data collected is often uncertainties.The fuzzy inference system can deal with the uncertainty information well and get a relatively good classification effect.The learning capabilities of the neural network which the samples with similar measurement values but different categories can be distinguished by adjusting the weights during the big data training process are not available in many other methods.However,the existing classification algorithms are more focused on the overall classification effect,while most of the actual data are distributed unevenly and the minority samples play a more important role.Traditional classifiers are more likely to be affected by the majority samples during model and ignore minority samples,which results in a poor classifier's effect on the classification of the minority samples,and the overall classification effect of classifiers will be affected to some extent.To solve the above problems and improve the performance of the fuzzy neural classifier,existing unbalanced data processing methods have been analysed and a hybrid unbalanced data processing algorithm based on data and algorithm level are proposed which not only solves the problem of data imbalance but also improves the classifier's effect on the classification of the minority in this paper.What' s more,the data processed by the method proposed is more in line with the characteristics of the original sample,and reduced sample noise,which can improve the overall classification accuracy of the classifier.To ensure lower computational complexity,the K-means clustering algorithm is used in the parameter generation stage,which not only reduces the computational complexity but also ensures better initial classification results.Finally,in order to make the recognition rate of the model reach a more ideal state,the particle swarm algorithm is used to optimize the system parameters which is a kind of global optimization algorithm that is not easy to fall into local optimum and while easy to be implemented,and UCI standard data is used to test the model proposed.The result proves that the method proposed in this paper reasonably processes the unbalanced data,and the accuracy of the model based on the proposed method is much higher than other methods.