HMM算法推导（附代码）

最新推荐文章于 2024-09-25 10:36:42 发布

yy031

最新推荐文章于 2024-09-25 10:36:42 发布

阅读量227

点赞数

分类专栏：笔记文章标签： HMM 维特比算法气象预测 emission_probability transition_probability

本文链接：https://blog.csdn.net/yy031/article/details/120538342

版权

笔记专栏收录该内容

4 篇文章 0 订阅

订阅专栏

推导过程：
在这里插入图片描述

完整代码（还有待改进）：



from numpy import *
import numpy as np

class HMM:

	def __init__(self):
		self.emission_probability=np.array([(0.1,0.4,0.5),(0.6,0.3,0.1)])
		self.transition_probability=np.array([(0.7,0.3),(0.4,0.6)])
		self.start_probability=np.array([(0.6),(0.4)])
		self.out=[1,0,0]
		self.t=len(self.out)
		self.m=len(self.emission_probability)
		self.n=len(self.transition_probability[0])


	def viterbi(self):
		self.z = array(zeros((self.t, self.m)))
		self.d = array(zeros((self.t, self.m)))
		for i in range(self.m):
			self.z[0][i]=self.start_probability[i]*self.transition_probability[i][self.out[0]]
			self.d[0][i]=0
		for j in range(1,self.t):
			for i in range(self.m):
				maxnum=self.z[j-1][0]*self.emission_probability[0][i]
				node=1
				for k in range(1,self.m):
					temp=self.z[j-1][k]*self.emission_probability[k][i]
					if(maxnum<temp):
						maxnum=temp
						node=k+1
				self.z[j][i]=maxnum*self.transition_probability[i][self.out[j]]
				self.d[j][i]=node
		max_probability=self.z[self.t-1][0]
		last_node=[1]
		temp=0
		for i in range(1,self.m):
			if(max_probability<self.z[self.t-1][i]):
				max_probability=self.z[self.t-1][i]
				last_node[0]=i+1
				temp=i
		i=self.t-1
		while(i>=1):
			last_node.append(self.d[i][temp])
			i=i-1
		temp=['o']
		temp[0]=int(last_node[len(last_node)-1])
		j=len(last_node)-2
		while j>=0:
			temp.append(int(last_node[j]))
			j=j-1
		print('1代表晴天，2代表雨天，则路径节点为')
		print(temp)
		print('该路径概率为'+str(max_probability))


if __name__ == '__main__':
	test=HMM()
	test.viterbi()