Classification Problems Based On Machine Learning

Data classification problem as an important research direction in the field of big data.its research method in recent years become more and more intelligent and tool.Classification problem based on machine learning is the research hotspot in recent years and will focus on the combination of Meta-heuristic optimization algorithms and artificial neural network method is becoming more and more widely used.its basic idea is: we use a Meta-heuristic optimization algorithms to adjustment the corresponding parameters of artificial neural network.they get the optimal value to be output after and then obtained parameters are substitution into neural networks for classification predictionThis paper mainly studies the classification of surface water quality and robot steering.We through the improvement of artificial neural network or combination Meta-heuristic optimization algorithms and machine learning,combining the experiment data to classify experiment.The experiment data are classified,and it is proved that the method proposed in this paper can achieve good results in accuracy.The paper from the practical point of view.It expounds the problems and common methods that are encountered in the classification of data.The main research content is using improved particle swarm optimization(PSO)to optimize the support vector machine(SVM),First,the basic principles of PSO and SVM are introduced.The advantages of PSO are high accuracy,fast convergence speed,SVM's universality,robustness,effectiveness and classification ability.The water quality of the surface water and the robot steering are classified and studied.The experimental results show that the improved algorithm has a higher classification accuracy.Secondly,an improved universal gravitational search algorithm(GSA)is proposed to optimize the BP neural network.Mainly using the improved algorithm to optimize weights and deviations in BP neural network,the optimization process is fully combined with PSO's global optimization and GSA's local search ability,and the exponential function is applied to the acceleration coefficient.The simulation results show that the improved algorithm can solve the problem of local minimum,and it has fast convergence speed.Finally,The S-kohonen neural network is introduced to classify the surface water quality and the robot steering.The S-kohonen neural network is to add an output layer and improve the weight value after joining the competitive layer on the basis of the Kohonen neural network.The experimental results are analyzed by studying the different number of iterations.The experimental results show that the greater the number of iterations,the more accurate the results.The results of the three methods of surface water quality and robot steering are compared and analyzed.The experimental results show that the I-PSO-GSA proposed in this paper has a very good classification accuracy.