每个神经网络对应每个收敛标准δ都有一个特征的迭代次数n,因此可以用迭代次数曲线n(δ)来评价网络性能。
一个二分类网络分类两个对象A和B,B中有K张图片,B的第i张图片被取样的概率为pi,B中第i张图片相对A的迭代次数为ni最终的迭代次数nt等于pi*ni的累加和。
由此可以构造两个矩阵一个是随机矩阵PJ
PJ表明图片集B中第i张图片被抽样到的概率
和矩阵NJ
NJ表明图片集B中第i张图片相对A的迭代次数
总的迭代次数nt等于矩阵PJ和NJ的点积
为了验证这个关系构造了等式
也就是B中有3张图片,让三张图片被抽样到的概率相等。
本文验算这个表达式是否正确
实验过程
首先用实验的方法测量n1
制作一个带一个3*3卷积核的神经网络,测试集是mnist的0和一张图片x,将28*28的图片缩小成9*9,隐藏层30个节点所以网络的结构是
这个网络分成两个部分左边的是让mnist 0向1,0收敛,右边的是让x向 0,1收敛。但是让左右两边的权重实现同步更新,实现权重共享。前面大量实验表明这种效果相当于将两个弹性系数为k1,k2的弹簧并联成一个弹性系数为k的弹簧,并且让k1=k2=k/2的过程。
将上图简写成
d2(mnist0, x=1)81-con(3*3)49-30-2-(2*k) ,k∈{0,1}
这个网络的收敛标准是
if (Math.abs(f2[0]-y[0])< δ && Math.abs(f2[1]-y[1])< δ )
本文尝试了δ从0.5到1e-6在内的26个值,训练集是mnist0
图片x就是一张二维数组,让x=1.
具体进样顺序 | ||||
进样顺序 | 迭代次数 | |||
δ=0.5 | ||||
mnist 0-1 | 1 | 判断是否达到收敛 | ||
X | 2 | 判断是否达到收敛 | ||
梯度下降 | ||||
mnist 0-2 | 3 | 判断是否达到收敛 | ||
X | 4 | 判断是否达到收敛 | ||
梯度下降 | ||||
…… | ||||
mnist 0-4999 | 9997 | 判断是否达到收敛 | ||
X | 9998 | 判断是否达到收敛 | ||
梯度下降 | ||||
…… | ||||
如果4999图片内没有达到收敛标准再次从头循环 | ||||
mnist 0-1 | 9999 | 判断是否达到收敛 | ||
X | 10000 | 判断是否达到收敛 | ||
…… | ||||
达到收敛标准记录迭代次数,将这个过程重复199次 | ||||
δ=0.4 | ||||
…… |
用这个方法可以得到网络
d2(mnist0, x=1)81-con(3*3)49-30-2-(2*k) ,k∈{0,1}
的迭代次数曲线n1。
第二步测量n0.6
用同样的办法制作另一个网络
d2(mnist0, x=0.6)81-con(3*3)49-30-2-(2*k) ,k∈{0,1}
让mnist 0向1,0收敛,右边的是让x向 0,1收敛。但让x=0.6.得到迭代次数曲线n0.6
第三步测量n0.1
用同样的办法制作另一个网络
d2(mnist0, x=0.1)81-con(3*3)49-30-2-(2*k) ,k∈{0,1}
让mnist 0向1,0收敛,右边的是让x向 0,1收敛。但让x=0.1.得到迭代次数曲线n0.1
实验数据
在《测量一组对角矩阵的频率和质量》中已经将这3个迭代次数都测出来了
1 | 0.6 | 0.1 | |
δ | 迭代次数n | 迭代次数n | 迭代次数n |
0.5 | 17.40201005 | 18.7839196 | 17.87437186 |
0.4 | 951.2110553 | 1117.532663 | 1408.577889 |
0.3 | 1144.577889 | 1415.271357 | 1720.517588 |
0.2 | 1313.633166 | 1623 | 1995.110553 |
0.1 | 1505.824121 | 1821.562814 | 2243.834171 |
0.01 | 2362.115578 | 2582.668342 | 3001.552764 |
0.001 | 4129.020101 | 3895.211055 | 4007.532663 |
1.00E-04 | 10353.37186 | 7057.452261 | 5532.668342 |
9.00E-05 | 10653.93467 | 7175.934673 | 5683.753769 |
8.00E-05 | 11292.43719 | 7392.497487 | 6131.934673 |
7.00E-05 | 11761.11055 | 8186.427136 | 6106.919598 |
6.00E-05 | 12657.69347 | 8620.758794 | 6014.688442 |
5.00E-05 | 13305.44221 | 9001.532663 | 6455.321608 |
4.00E-05 | 15844.29648 | 9694.778894 | 6724.738693 |
3.00E-05 | 17291.77387 | 10665.48241 | 7055.80402 |
2.00E-05 | 20753.56281 | 11822.9397 | 7763.41206 |
1.00E-05 | 27708.19598 | 15788.16583 | 8749.050251 |
9.00E-06 | 29358.8593 | 15766.86935 | 8879.41206 |
8.00E-06 | 30689.87437 | 16904.54774 | 9387.150754 |
7.00E-06 | 33437.22111 | 18446.72864 | 9532.648241 |
6.00E-06 | 36960.63819 | 18505.60302 | 9957.683417 |
5.00E-06 | 40669.92462 | 19916.76884 | 10661.56281 |
4.00E-06 | 44594.04523 | 20718.10553 | 11025.0402 |
3.00E-06 | 51522.10553 | 26158.03518 | 11653.63317 |
2.00E-06 | 67583.53266 | 26274.55779 | 13076.9196 |
1.00E-06 | 107224.5276 | 37806.51759 | 15184.58794 |
现在做第4个网络
d2(mnist0 ; 33.33% x=1, 33.33%x=0.6,33.33%x=0.1)81-con(3*3)49-30-2-(2*k) ,k∈{0,1}
让mnist 0向1,0收敛,右边的是让x向 0,1收敛。但让x在1,0.6,0.1之间随机。
让1:0.6:0.1的比例是1:1:1.
具体进样顺序 |
| |||
进样顺序 | 迭代次数 |
| ||
δ=0.5 |
| |||
mnist 0-1 | 1 | 判断是否达到收敛 | ||
33.33% x=1,33.33% x=0.6,33.33% x=0.1 | 2 | 判断是否达到收敛 | ||
梯度下降 |
| |||
mnist 0-2 | 3 | 判断是否达到收敛 | ||
33.33% x=1,33.33% x=0.6,33.33% x=0.1 | 4 | 判断是否达到收敛 | ||
梯度下降 |
| |||
…… |
| |||
mnist 0-4999 | 9997 | 判断是否达到收敛 | ||
33.33% x=1,33.33% x=0.6,33.33% x=0.1 | 9998 | 判断是否达到收敛 | ||
梯度下降 |
| |||
…… |
| |||
如果4999图片内没有达到收敛标准再次从头循环 |
| |||
mnist 0-1 | 9999 | 判断是否达到收敛 | ||
33.33% x=1,33.33% x=0.6,33.33% x=0.1 | 10000 | 判断是否达到收敛 | ||
…… |
| |||
达到收敛标准记录迭代次数,将这个过程重复199次 |
| |||
δ=0.4 |
| |||
…… |
|
相当于分类两个图片集,一个图片集是mnist的0另一个图片集只有三张图片,三张图片被取样的概率是1:1:1
得到的数据
用0和x二分类 | ||||||||
1:0.6:0.1=1:1:1 | ||||||||
f2[0] | f2[1] | 迭代次数n | 平均准确率p-ave | δ | 耗时ms/次 | 耗时ms/199次 | 耗时min/199次 | 最大准确率p-max |
0.500154925 | 0.501585276 | 16.4321608 | 0.505957684 | 0.5 | 794.9346734 | 158208 | 2.6368 | 0.781560284 |
0.609011221 | 0.391362551 | 1127.552764 | 0.481050643 | 0.4 | 974.9346734 | 194028 | 3.2338 | 0.901654846 |
0.714350091 | 0.285408928 | 1397.331658 | 0.58525488 | 0.3 | 1020.864322 | 203152 | 3.385866667 | 0.995744681 |
0.815157448 | 0.184737316 | 1581 | 0.645950794 | 0.2 | 1053.824121 | 209711 | 3.495183333 | 0.99858156 |
0.913633704 | 0.086425773 | 1778.035176 | 0.651574658 | 0.1 | 1088.075377 | 216527 | 3.608783333 | 0.997163121 |
0.992158629 | 0.007834828 | 2633.482412 | 0.661171104 | 0.01 | 1226.879397 | 244165 | 4.069416667 | 0.997635934 |
0.999259688 | 7.40E-04 | 4070.829146 | 0.593886691 | 0.001 | 1465.135678 | 291562 | 4.859366667 | 0.998108747 |
0.999926055 | 7.39E-05 | 7549.361809 | 0.561754398 | 1.00E-04 | 2041.457286 | 406266 | 6.7711 | 0.994326241 |
0.999932673 | 6.71E-05 | 8045.884422 | 0.568585243 | 9.00E-05 | 2119.165829 | 421716 | 7.0286 | 0.997163121 |
0.999940845 | 5.92E-05 | 8205.291457 | 0.552469202 | 8.00E-05 | 2138.552764 | 425577 | 7.09295 | 0.997163121 |
0.999945526 | 5.44E-05 | 8975.321608 | 0.553949416 | 7.00E-05 | 2274.859296 | 452699 | 7.544983333 | 0.994326241 |
0.999952729 | 4.73E-05 | 9572.648241 | 0.548662936 | 6.00E-05 | 2020.763819 | 402136 | 6.702266667 | 0.998108747 |
0.999962313 | 3.76E-05 | 9745.703518 | 0.549793887 | 5.00E-05 | 2460.839196 | 489709 | 8.161816667 | 0.987234043 |
0.999969451 | 3.05E-05 | 10831.36181 | 0.551411906 | 4.00E-05 | 2634.407035 | 524248 | 8.737466667 | 0.995744681 |
0.999977419 | 2.26E-05 | 12239.86432 | 0.545982869 | 3.00E-05 | 2829.522613 | 563076 | 9.3846 | 0.996690307 |
0.999985136 | 1.49E-05 | 12683.34171 | 0.568404671 | 2.00E-05 | 2937.442211 | 584551 | 9.742516667 | 0.998108747 |
0.999992339 | 7.66E-06 | 17797.66332 | 0.564460601 | 1.00E-05 | 3788.673367 | 753947 | 12.56578333 | 0.997635934 |
0.999993191 | 6.82E-06 | 18891.29146 | 0.553564513 | 9.00E-06 | 3421.296482 | 680854 | 11.34756667 | 0.996690307 |
0.999993866 | 6.12E-06 | 18053.13065 | 0.568870356 | 8.00E-06 | 3813.798995 | 758946 | 12.6491 | 0.997635934 |
0.999994712 | 5.29E-06 | 20724.70854 | 0.531960037 | 7.00E-06 | 4226.748744 | 841139 | 14.01898333 | 0.992434988 |
0.999995376 | 4.62E-06 | 22732.40704 | 0.538610309 | 6.00E-06 | 4630.934673 | 921556 | 15.35926667 | 0.997635934 |
0.999996217 | 3.78E-06 | 23539.19095 | 0.540104779 | 5.00E-06 | 4390.59799 | 873737 | 14.56228333 | 0.994326241 |
0.999996903 | 3.10E-06 | 26331.90452 | 0.530458439 | 4.00E-06 | 5423.693467 | 1079331 | 17.98885 | 0.996690307 |
0.999997724 | 2.27E-06 | 30580.90955 | 0.543281419 | 3.00E-06 | 5887.869347 | 1171718 | 19.52863333 | 0.996690307 |
0.999998466 | 1.54E-06 | 32664.11558 | 0.544578685 | 2.00E-06 | 6201.030151 | 1234005 | 20.56675 | 0.992907801 |
0.999999193 | 8.05E-07 | 54834.03518 | 0.525661404 | 1.00E-06 | 9782.688442 | 1946756 | 32.44593333 | 0.989598109 |
所以现在有了4个迭代次数分别是
x=1 | n1 |
x=0.6 | n0.6 |
x=0.1 | n0.1 |
0.3333x=1||0.3333x=0.6||0.3333x=0.1 | n1-0.6-0.1 |
验算这4个值之间的关系
1 | 0.6 | 0.1 | 理论值 | 实测值 | 理论值/实测值 | |
δ | 迭代次数n | 迭代次数n | 迭代次数n | (n0.1+n0.6+n1)/3 |
|
|
0.5 | 17.40201005 | 18.7839196 | 17.87437186 | 18.0201005 | 16.4321608 | 1.096636086 |
0.4 | 951.2110553 | 1117.532663 | 1408.577889 | 1159.107203 | 1127.552764 | 1.027984889 |
0.3 | 1144.577889 | 1415.271357 | 1720.517588 | 1426.788945 | 1397.331658 | 1.021081099 |
0.2 | 1313.633166 | 1623 | 1995.110553 | 1643.914573 | 1581 | 1.039794164 |
0.1 | 1505.824121 | 1821.562814 | 2243.834171 | 1857.073702 | 1778.035176 | 1.044452735 |
0.01 | 2362.115578 | 2582.668342 | 3001.552764 | 2648.778894 | 2633.482412 | 1.005808462 |
0.001 | 4129.020101 | 3895.211055 | 4007.532663 | 4010.58794 | 4070.829146 | 0.985201736 |
1.00E-04 | 10353.37186 | 7057.452261 | 5532.668342 | 7647.830821 | 7549.361809 | 1.013043356 |
9.00E-05 | 10653.93467 | 7175.934673 | 5683.753769 | 7837.874372 | 8045.884422 | 0.974147025 |
8.00E-05 | 11292.43719 | 7392.497487 | 6131.934673 | 8272.289782 | 8205.291457 | 1.008165258 |
7.00E-05 | 11761.11055 | 8186.427136 | 6106.919598 | 8684.819095 | 8975.321608 | 0.967633192 |
6.00E-05 | 12657.69347 | 8620.758794 | 6014.688442 | 9097.713568 | 9572.648241 | 0.950386282 |
5.00E-05 | 13305.44221 | 9001.532663 | 6455.321608 | 9587.432161 | 9745.703518 | 0.983759884 |
4.00E-05 | 15844.29648 | 9694.778894 | 6724.738693 | 10754.60469 | 10831.36181 | 0.992913438 |
3.00E-05 | 17291.77387 | 10665.48241 | 7055.80402 | 11671.0201 | 12239.86432 | 0.953525284 |
2.00E-05 | 20753.56281 | 11822.9397 | 7763.41206 | 13446.63819 | 12683.34171 | 1.060181023 |
1.00E-05 | 27708.19598 | 15788.16583 | 8749.050251 | 17415.13735 | 17797.66332 | 0.978506956 |
9.00E-06 | 29358.8593 | 15766.86935 | 8879.41206 | 18001.71357 | 18891.29146 | 0.952910689 |
8.00E-06 | 30689.87437 | 16904.54774 | 9387.150754 | 18993.85762 | 18053.13065 | 1.0521088 |
7.00E-06 | 33437.22111 | 18446.72864 | 9532.648241 | 20472.19933 | 20724.70854 | 0.987816031 |
6.00E-06 | 36960.63819 | 18505.60302 | 9957.683417 | 21807.97487 | 22732.40704 | 0.959334172 |
5.00E-06 | 40669.92462 | 19916.76884 | 10661.56281 | 23749.41876 | 23539.19095 | 1.00893097 |
4.00E-06 | 44594.04523 | 20718.10553 | 11025.0402 | 25445.73032 | 26331.90452 | 0.966345989 |
3.00E-06 | 51522.10553 | 26158.03518 | 11653.63317 | 29777.92462 | 30580.90955 | 0.973742281 |
2.00E-06 | 67583.53266 | 26274.55779 | 13076.9196 | 35645.00335 | 32664.11558 | 1.091258793 |
1.00E-06 | 107224.5276 | 37806.51759 | 15184.58794 | 53405.21106 | 54834.03518 | 0.973942751 |
从数值看
这个公式还是符合的很好的。
表明神经网络的迭代次数可以被看作是一个线性变量可以用概率矩阵和迭代次数矩阵的点积来计算。
实验参数
学习率 0.1 |
权重初始化方式 |
Random rand1 =new Random(); |
int ti1=rand1.nextInt(98)+1; |
int xx=1; |
if(ti1%2==0) |
{ xx=-1;} |
tw[a][b]=xx*((double)ti1/x); |
第一层第二层和卷积核的权重的初始化的x分别为1000,1000,200 |
0.1 | ||||||||
f2[0] | f2[1] | 迭代次数n | 平均准确率p-ave | δ | 耗时ms/次 | 耗时ms/199次 | 耗时min/199次 | 最大准确率p-max |
0.498913 | 0.4988 | 17.87437186 | 0.516198011 | 0.5 | 711.7789 | 141644 | 2.360733 | 0.980142 |
0.609699 | 0.390362 | 1408.577889 | 0.526873136 | 0.4 | 936.4472 | 186353 | 3.105883 | 0.992908 |
0.714834 | 0.28527 | 1720.517588 | 0.681069651 | 0.3 | 1005.05 | 200008 | 3.333467 | 0.997636 |
0.816705 | 0.182972 | 1995.110553 | 0.806410302 | 0.2 | 1030.407 | 205058 | 3.417633 | 0.997163 |
0.914477 | 0.085512 | 2243.834171 | 0.815267829 | 0.1 | 1073.216 | 213575 | 3.559583 | 0.997163 |
0.99262 | 0.007391 | 3001.552764 | 0.772726517 | 0.01 | 1195.271 | 237867 | 3.96445 | 0.996217 |
0.999292 | 7.08E-04 | 4007.532663 | 0.699901398 | 0.001 | 1362.05 | 271052 | 4.517533 | 0.996217 |
0.999931 | 6.92E-05 | 5532.668342 | 0.670097533 | 1.00E-04 | 1613.925 | 321179 | 5.352983 | 0.997163 |
0.999941 | 5.85E-05 | 5683.753769 | 0.624410468 | 9.00E-05 | 1638.111 | 325989 | 5.43315 | 0.998109 |
0.999948 | 5.21E-05 | 6131.934673 | 0.628342659 | 8.00E-05 | 1712.166 | 340723 | 5.678717 | 0.998109 |
0.999953 | 4.72E-05 | 6106.919598 | 0.621794552 | 7.00E-05 | 1747.035 | 347675 | 5.794583 | 0.997163 |
0.999961 | 3.86E-05 | 6014.688442 | 0.660669779 | 6.00E-05 | 1731.799 | 344628 | 5.7438 | 0.997163 |
0.999967 | 3.31E-05 | 6455.321608 | 0.644967153 | 5.00E-05 | 1799.472 | 358095 | 5.96825 | 0.998109 |
0.999973 | 2.68E-05 | 6724.738693 | 0.620573316 | 4.00E-05 | 1847.055 | 367609 | 6.126817 | 0.997636 |
0.99998 | 1.96E-05 | 7055.80402 | 0.640904285 | 3.00E-05 | 1904.673 | 379030 | 6.317167 | 0.997636 |
0.999986 | 1.36E-05 | 7763.41206 | 0.63548 | 2.00E-05 | 2063.276 | 410593 | 6.843217 | 0.997636 |
0.999993 | 6.88E-06 | 8749.050251 | 0.621000986 | 1.00E-05 | 2808.357 | 558880 | 9.314667 | 0.997636 |
0.999994 | 6.20E-06 | 8879.41206 | 0.624534018 | 9.00E-06 | 2812.92 | 559771 | 9.329517 | 0.99669 |
0.999994 | 5.60E-06 | 9387.150754 | 0.632904475 | 8.00E-06 | 2287.397 | 455224 | 7.587067 | 0.997636 |
0.999995 | 4.59E-06 | 9532.648241 | 0.609351723 | 7.00E-06 | 2942.317 | 585537 | 9.75895 | 0.997163 |
0.999996 | 3.95E-06 | 9957.683417 | 0.618418333 | 6.00E-06 | 3023.276 | 601648 | 10.02747 | 0.99669 |
0.999997 | 3.33E-06 | 10661.56281 | 0.585715813 | 5.00E-06 | 3185.151 | 633860 | 10.56433 | 0.996217 |
0.999997 | 2.67E-06 | 11025.0402 | 0.613820877 | 4.00E-06 | 3263.668 | 649470 | 10.8245 | 0.998109 |
0.999998 | 1.97E-06 | 11653.63317 | 0.59831783 | 3.00E-06 | 3388.819 | 674375 | 11.23958 | 0.998109 |
0.999999 | 1.40E-06 | 13076.9196 | 0.631728382 | 2.00E-06 | 3393.688 | 675344 | 11.25573 | 0.998109 |
0.999999 | 6.81E-07 | 15184.58794 | 0.620981978 | 1.00E-06 | 4139.754 | 823826 | 13.73043 | 0.997163 |
xx0 | ||||||||
0.6 | ||||||||
f2[0] | f2[1] | 迭代次数n | 平均准确率p-ave | δ | 耗时ms/次 | 耗时ms/199次 | 耗时min/199次 | 最大准确率p-max |
0.501251 | 0.496685 | 18.78392 | 0.502121 | 0.5 | 690.3518 | 137392 | 2.289867 | 0.864303 |
0.608328 | 0.391587 | 1117.533 | 0.477078 | 0.4 | 528.3317 | 105146 | 1.752433 | 0.913948 |
0.71453 | 0.285302 | 1415.271 | 0.548513 | 0.3 | 902.2764 | 179560 | 2.992667 | 0.976832 |
0.814621 | 0.185598 | 1623 | 0.621562 | 0.2 | 935.6935 | 186212 | 3.103533 | 0.997636 |
0.913459 | 0.086546 | 1821.563 | 0.626936 | 0.1 | 968.6683 | 192770 | 3.212833 | 0.997163 |
0.992169 | 0.007831 | 2582.668 | 0.638065 | 0.01 | 1091.97 | 217308 | 3.6218 | 0.994326 |
0.999265 | 7.34E-04 | 3895.211 | 0.57965 | 0.001 | 1306.281 | 259954 | 4.332567 | 0.998582 |
0.999925 | 7.46E-05 | 7057.452 | 0.535956 | 1.00E-04 | 1823.095 | 362802 | 6.0467 | 0.993853 |
0.999934 | 6.60E-05 | 7175.935 | 0.543583 | 9.00E-05 | 1842.191 | 366601 | 6.110017 | 0.997636 |
0.999939 | 6.11E-05 | 7392.497 | 0.546869 | 8.00E-05 | 1881.804 | 374488 | 6.241467 | 0.996217 |
0.999947 | 5.26E-05 | 8186.427 | 0.539777 | 7.00E-05 | 2011.653 | 400326 | 6.6721 | 0.998109 |
0.999954 | 4.58E-05 | 8620.759 | 0.521677 | 6.00E-05 | 2157.844 | 429422 | 7.157033 | 0.937589 |
0.999961 | 3.88E-05 | 9001.533 | 0.541614 | 5.00E-05 | 2163.774 | 430596 | 7.1766 | 0.992435 |
0.999969 | 3.07E-05 | 9694.779 | 0.541671 | 4.00E-05 | 2323.246 | 462333 | 7.70555 | 0.997636 |
0.999976 | 2.41E-05 | 10665.48 | 0.538734 | 3.00E-05 | 2698.648 | 537042 | 8.9507 | 0.994326 |
0.999984 | 1.57E-05 | 11822.94 | 0.559813 | 2.00E-05 | 2962.558 | 589557 | 9.82595 | 0.987707 |
0.999992 | 7.76E-06 | 15788.17 | 0.537325 | 1.00E-05 | 3709.754 | 738261 | 12.30435 | 0.986288 |
0.999993 | 7.02E-06 | 15766.87 | 0.54288 | 9.00E-06 | 3707.593 | 737811 | 12.29685 | 0.994799 |
0.999994 | 6.14E-06 | 16904.55 | 0.526391 | 8.00E-06 | 3835.543 | 763281 | 12.72135 | 0.993853 |
0.999995 | 5.43E-06 | 18446.73 | 0.510261 | 7.00E-06 | 4095.553 | 815031 | 13.58385 | 0.997163 |
0.999995 | 4.71E-06 | 18505.6 | 0.534355 | 6.00E-06 | 4101.186 | 816137 | 13.60228 | 0.991017 |
0.999996 | 3.93E-06 | 19916.77 | 0.527667 | 5.00E-06 | 4115.08 | 818901 | 13.64835 | 0.987234 |
0.999997 | 3.13E-06 | 20718.11 | 0.532616 | 4.00E-06 | 4503.146 | 896126 | 14.93543 | 0.983452 |
0.999998 | 2.41E-06 | 26158.04 | 0.52708 | 3.00E-06 | 5511.734 | 1096851 | 18.28085 | 0.994326 |
0.999998 | 1.53E-06 | 26274.56 | 0.524317 | 2.00E-06 | 5363.884 | 1067430 | 17.7905 | 0.985343 |
0.999999 | 7.88E-07 | 37806.52 | 0.510657 | 1.00E-06 | 7611.156 | 1514620 | 25.24367 | 0.995272 |
xx0 | ||||||||
1 | ||||||||
f2[0] | f2[1] | 迭代次数n | 平均准确率p-ave | δ | 耗时ms/次 | 耗时ms/199次 | 耗时min/199次 | 最大准确率p-max |
0.499717 | 0.500211 | 17.40201 | 0.506172 | 0.5 | 707.1759 | 140728 | 2.345467 | 0.838298 |
0.608011 | 0.392067 | 951.2111 | 0.469311 | 0.4 | 850.4422 | 169238 | 2.820633 | 0.791962 |
0.712017 | 0.287774 | 1144.578 | 0.487038 | 0.3 | 920.7638 | 183234 | 3.0539 | 0.960284 |
0.813526 | 0.186626 | 1313.633 | 0.519771 | 0.2 | 921.196 | 183322 | 3.055367 | 0.995272 |
0.910566 | 0.089368 | 1505.824 | 0.523751 | 0.1 | 955.1608 | 190081 | 3.168017 | 0.992435 |
0.991599 | 0.008414 | 2362.116 | 0.518659 | 0.01 | 1100.508 | 219019 | 3.650317 | 0.995745 |
0.999177 | 8.23E-04 | 4129.02 | 0.515074 | 0.001 | 1356.719 | 270003 | 4.50005 | 0.995272 |
0.999915 | 8.50E-05 | 10353.37 | 0.488022 | 1.00E-04 | 2390.417 | 475693 | 7.928217 | 0.949882 |
0.999923 | 7.65E-05 | 10653.93 | 0.483514 | 9.00E-05 | 2434.392 | 484444 | 8.074067 | 0.968322 |
0.999932 | 6.78E-05 | 11292.44 | 0.487855 | 8.00E-05 | 2542.327 | 505923 | 8.43205 | 0.982979 |
0.999943 | 5.74E-05 | 11761.11 | 0.486713 | 7.00E-05 | 2613.879 | 520162 | 8.669367 | 0.904019 |
0.999949 | 5.14E-05 | 12657.69 | 0.49194 | 6.00E-05 | 2541.392 | 505737 | 8.42895 | 0.908274 |
0.999958 | 4.15E-05 | 13305.44 | 0.484669 | 5.00E-05 | 2879.116 | 572945 | 9.549083 | 0.992908 |
0.999967 | 3.35E-05 | 15844.3 | 0.488837 | 4.00E-05 | 3299.538 | 656624 | 10.94373 | 0.93617 |
0.999975 | 2.50E-05 | 17291.77 | 0.494123 | 3.00E-05 | 3521.02 | 700698 | 11.6783 | 0.983924 |
0.999983 | 1.69E-05 | 20753.56 | 0.486458 | 2.00E-05 | 4093.045 | 814516 | 13.57527 | 0.946572 |
0.999992 | 8.35E-06 | 27708.2 | 0.485498 | 1.00E-05 | 5227.101 | 1040193 | 17.33655 | 0.982506 |
0.999992 | 7.53E-06 | 29358.86 | 0.484006 | 9.00E-06 | 5562.97 | 1107046 | 18.45077 | 0.901182 |
0.999993 | 6.75E-06 | 30689.87 | 0.488079 | 8.00E-06 | 5993.101 | 1192630 | 19.87717 | 0.947991 |
0.999994 | 5.85E-06 | 33437.22 | 0.478233 | 7.00E-06 | 5531.07 | 1100684 | 18.34473 | 0.922459 |
0.999995 | 5.10E-06 | 36960.64 | 0.476242 | 6.00E-06 | 6762.432 | 1345724 | 22.42873 | 0.839243 |
0.999996 | 4.27E-06 | 40669.92 | 0.479425 | 5.00E-06 | 7354.492 | 1463560 | 24.39267 | 0.953191 |
0.999997 | 3.40E-06 | 44594.05 | 0.486247 | 4.00E-06 | 8140.05 | 1619870 | 26.99783 | 0.97305 |
0.999997 | 2.58E-06 | 51522.11 | 0.482569 | 3.00E-06 | 9523.98 | 1895272 | 31.58787 | 0.975414 |
0.999998 | 1.77E-06 | 67583.53 | 0.474122 | 2.00E-06 | 12034.15 | 2394797 | 39.91328 | 0.985816 |
0.999999 | 9.08E-07 | 107224.5 | 0.472789 | 1.00E-06 | 18710.25 | 3723339 | 62.05565 | 0.886998 |