粒子群算法优化BP神经网络

来源：互联网发布：linux 监听端口编辑：程序博客网时间：2024/05/17 03:14

刚学粒子群算法，然后用粒子群算法优化神经网络的隐藏节点数，代码写的不是太好，如果代码有问题，请大家多多指教。第一次写博客，多多包涵。

本文的粒子群算法用的是标准粒子群算法，权重更新采用线性递减策略。

这是SSE图

import numpy as npimport random import matplotlib.pyplot as pltimport neurolab as nltrain_x =[]d=[]samplescount=1000myrndsmp=np.random.rand(samplescount)for yb_i in xrange(0,samplescount):    train_x.append([myrndsmp[yb_i]*4*np.pi-2*np.pi])for yb_i in xrange(0,samplescount):    d.append(np.sin(train_x[yb_i])*0.5+np.cos(train_x[yb_i])*0.5)myinput=np.array(train_x)   mytarget=np.array(d)#PSO参数设置class PSO():    def __init__(self,max_iter):        #self.w = 0.8          self.c1 = 2           self.c2 = 2           self.pN =10               #粒子数量        self.dim = 1              #搜索维度        self.max_iter = max_iter    #迭代次数        self.X = np.ones((self.pN,self.dim))       #所有粒子的位置和速度        self.V = np.zeros((self.pN,self.dim))        self.pbest = np.zeros((self.pN,self.dim))   #个体经历的最佳位置和全局最佳位置        self.gbest = np.zeros((1,self.dim))        self.p_fit = np.zeros(self.pN)              #每个个体的历史最佳适应值        self.fit = 1e10             #全局最佳适应值        self.wmax=0.9        self.wmin=0.4#目标函数    def fun(self,err):        fitness=err        return fitness#初始化种群    def init_Population(self):        for i in range(self.pN):            for j in range(self.dim):                self.X[i][j] = random.uniform(1,20)                self.V[i][j] = random.uniform(0,2)            self.pbest[i] = self.X[i]                      for x in self.pbest[i]:                return x            bpnet = nl.net.newff([[-2*np.pi, 2*np.pi]], [int(x)+1, 1])            err = bpnet.train(myinput, mytarget, epochs=800, show=100, goal=0.02)            #out=net.sim(input)            tmp = self.fun(err)            self.p_fit[i] = tmp            if(tmp < self.fit):                self.fit = tmp                self.gbest = self.X[i]    #更新粒子位置    def iterator(self):        fitness = []        for t in range(self.max_iter):            w=self.wmax-(self.wmax-self.wmin)*(float(t)/self.max_iter)            for i in range(self.pN):                for x in self.pbest[i]:                    return x                print x                bpnet = nl.net.newff([[-2*np.pi, 2*np.pi]], [int(x)+1, 1])                err = bpnet.train(myinput, mytarget, epochs=800, show=100, goal=0.02)                temp = self.fun(err)                if(temp

阅读全文

1 0