Generate Possion Dis.
step1:Generate a random number between [0,1]
step2:Let u=F(x)=1-[(1/e)x]
step3:Slove x=1/F(u)
step4:Repeat Step1~Step3 by using different u,you can get x1,x2,x3,...,xn
step5:If the first packet was generated at time [0], than the
second packet will be generated at time [0+x1],The third packet will be generated at time [0+x1+x2],
and so on ….
Random-number generation
1.static method random from class Math
-Returns doubles in the range 0.0 <= x < 1.0
2.class Random from package java.util
-Can produce pseudorandom boolean, byte, float, double, int, long and Gaussian values
-Is seeded with the current time of day to generate different sequences of numbers each time the program executes
標(biāo)簽:
Generate
Possion
between
random
上傳時間:
2017-05-25
上傳用戶:bibirnovis
提出了一種用于矢量量化的改進(jìn)的聚類算法,該算法在MKM(Modified K-Means)算法的框架的基礎(chǔ)上,對初始碼本的生成、失真測度的選擇、非典型胞腔的處理等方面進(jìn)行了改進(jìn),從而減少了原算法在能量和增益上對聚類結(jié)果的影響.并將該算法應(yīng)用于波形編輯孤立字識別器,這種識別器直接對語音樣本的時域波形進(jìn)行訓(xùn)練和聚類,不需要提取語音參數(shù),算法復(fù)雜度較低,加上提出的聚類算法失真測度簡單易實(shí)現(xiàn),對芯片的運(yùn)算能力要求不高,非常適用于有低成本要求的語音識別器場合.通過中文元音字識別的實(shí)驗(yàn)證明,在相同碼本尺寸下,運(yùn)用改進(jìn)后的聚類算法的識別器的識別率有所提高.
標(biāo)簽:
Modified
K-Means
算法
MKM
上傳時間:
2017-05-30
上傳用戶:tianjinfan