输入对5层网络迭代次数的影响
制作一个5层网络和一个3层网络
图中左边的5层网络很显然可以看作是由两个右边的3层网络组合而成，所以左边的网络的迭代次数和右边的网络的迭代次数有什么关系？
在《测量一组5层网络的迭代次数》测量了输入固定为0.1时5层网络和3层网络收敛迭代次数之间的关系。本文测量当输入是0-1之间的随机数时对迭代次数的影响。
将左边的5层网络写成
（r）-3*6*3*6*3-(3*k)，k∈{0,1}
意思是向网络输入0到1的随机数r，输出是1，0，0。
右边的3层网络写成
（r）-3*6*3-(3*k)，k∈{0,1}
其中r的初始化方法是
Random rand1 =new Random();
int ti1=rand1.nextInt(98)+1;
r=((double)ti1/100);
用这种办法制作了三组网络
（r）-2-10-2-10-2-（2*k），k∈{0,1}
（r）-2-10-2-（2*k），k∈{0,1}
（r）-3-10-3-10-3-（3*k），k∈{0,1}
（r）-3-10-3-（3*k），k∈{0,1}
（r）-4-10-4-10-4-（4*k），k∈{0,1}
（r）-4-10-4-（4*k），k∈{0,1}
用来比较
（r）-x-10-x-10-x-（x*k），k∈{0,1}
（r）-x-10-x-（x*k），k∈{0,1}
这两类网络迭代次数的关系。
具体的实验过程
网络的收敛标准是
if (Math.abs(f4[0]-y[0])< δ && Math.abs(f4[1]-y[1])< δ && Math.abs(f4[2]-y[2])< δ )
因为对应每个收敛标准δ都有一个特征的迭代次数n与之对应因此可以用迭代次数曲线n(δ)来评价网络性能。
|
具体进样顺序 |
|||
|
δ=0.5 |
迭代次数 |
||
|
r |
1 |
判断是否达到收敛 |
|
|
梯度下降 |
|||
|
r |
2 |
判断是否达到收敛 |
|
|
梯度下降 |
|||
|
…… |
|||
|
每当网路达到收敛标准记录迭代次数 |
|||
|
将这一过程重复199次 |
|||
|
δ=0.4 |
|||
|
…… |
|||
|
δ=1e-7 |
|||
本文尝试了δ从1e-7到0.5的共35个值，当网络满足条件收敛是记录迭代次数
首先观察5层网络迭代次数的变化
|
r-2-10-2-10-2 |
r-3-10-3-10-3 |
r-4-10-4-10-4 |
0.1-2-10-2-10-2 |
0.1-3-10-3-10-3 |
0.1-4-10-4-10-4 |
|||||||||||
|
δ |
迭代次数n |
迭代次数n |
迭代次数n |
r2/r2 |
r3/r2 |
r4/r2 |
δ |
迭代次数n |
迭代次数n |
迭代次数n |
0.1-2/0.1-2 |
0.1-3/0.12 |
0.1-4/0.1-2 |
r-2/0.1-2 |
r-3/0.1-3 |
r-4/0.1-4 |
|
0.5 |
1.904522613 |
1.869346734 |
3 |
1 |
0.981530343 |
1.575197889 |
0.5 |
1.934673367 |
1.959798995 |
3 |
1 |
1.012987013 |
1.550649351 |
0.984415584 |
0.953846154 |
1 |
|
0.4 |
5.341708543 |
5.331658291 |
7 |
1 |
0.998118532 |
1.310442145 |
0.4 |
5.371859296 |
5.341708543 |
7 |
1 |
0.994387278 |
1.303086997 |
0.994387278 |
0.998118532 |
1 |
|
0.3 |
10.34673367 |
10.22613065 |
12 |
1 |
0.988343856 |
1.159786304 |
0.3 |
10.27135678 |
10.18090452 |
11 |
1 |
0.991193738 |
1.070939335 |
1.007338552 |
1.004442251 |
1.090909091 |
|
0.2 |
18.70854271 |
18.30653266 |
20 |
1 |
0.978511953 |
1.069030352 |
0.2 |
18.66331658 |
18.52763819 |
20 |
1 |
0.99273021 |
1.071620894 |
1.002423263 |
0.988066178 |
1 |
|
0.1 |
40.25125628 |
38.37688442 |
38 |
1 |
0.953433208 |
0.944069913 |
0.1 |
40.10050251 |
38.36683417 |
39 |
1 |
0.956766917 |
0.972556391 |
1.003759399 |
1.000261952 |
0.974358974 |
|
0.01 |
325.8040201 |
268.6934673 |
235 |
1 |
0.824708876 |
0.721292512 |
0.01 |
325.6130653 |
268.5025126 |
237 |
1 |
0.824606078 |
0.727857771 |
1.000586447 |
1.000711184 |
0.991561181 |
|
0.001 |
2279.577889 |
1680.497487 |
1398 |
1 |
0.737196783 |
0.613271434 |
0.001 |
2280.115578 |
1684.38191 |
1406 |
1 |
0.738726548 |
0.616635408 |
0.999764183 |
0.997693859 |
0.9943101 |
|
1.00E-04 |
15912.24623 |
12004.45729 |
10590 |
1 |
0.75441626 |
0.665525146 |
1.00E-04 |
15835.47739 |
12056.8191 |
10609 |
1 |
0.76138021 |
0.669951384 |
1.004847902 |
0.995657079 |
0.998209068 |
|
9.00E-05 |
17423.96482 |
13189.36181 |
11673 |
1 |
0.756966738 |
0.669939369 |
9.00E-05 |
17342.0603 |
13253.43719 |
11717 |
1 |
0.764236599 |
0.675640598 |
1.004722883 |
0.995165376 |
0.996244773 |
|
8.00E-05 |
19282.14573 |
14663.1608 |
13043 |
1 |
0.760452753 |
0.676428868 |
8.00E-05 |
19208.1206 |
14739.47739 |
13077 |
1 |
0.767356562 |
0.680805805 |
1.003853846 |
0.9948223 |
0.997400015 |
|
7.00E-05 |
21687.30653 |
16544.53266 |
14783 |
1 |
0.762867101 |
0.681642968 |
7.00E-05 |
21562.95477 |
16633.47236 |
14844 |
1 |
0.771391145 |
0.688402872 |
1.005766917 |
0.994652969 |
0.995890596 |
|
6.00E-05 |
24788.30151 |
19034.80402 |
17128 |
1 |
0.767894646 |
0.690971102 |
6.00E-05 |
24676.82412 |
19138.08543 |
17197 |
1 |
0.775548966 |
0.696888705 |
1.004517493 |
0.994603357 |
0.995987672 |
|
5.00E-05 |
29117.86935 |
22492.8392 |
20378 |
1 |
0.772475449 |
0.699845162 |
5.00E-05 |
28974.85427 |
22613.09045 |
20480 |
1 |
0.780438453 |
0.706819776 |
1.004935834 |
0.994682228 |
0.995019531 |
|
4.00E-05 |
35437.03015 |
27629.55779 |
25185 |
1 |
0.779680399 |
0.710697254 |
4.00E-05 |
35289.97487 |
27783.49749 |
25306 |
1 |
0.787291507 |
0.717087504 |
1.004167055 |
0.994459312 |
0.995218525 |
|
3.00E-05 |
45809.18593 |
36073.57286 |
33209 |
1 |
0.787474654 |
0.724941937 |
3.00E-05 |
45565.71357 |
36295.21106 |
33318 |
1 |
0.796546531 |
0.731207686 |
1.005343324 |
0.993893459 |
0.996728495 |
|
2.00E-05 |
65988.56784 |
52783.36181 |
49229 |
1 |
0.799886458 |
0.746023161 |
2.00E-05 |
65556.1005 |
53122.43719 |
49468 |
1 |
0.810335526 |
0.754590337 |
1.006596905 |
0.993617097 |
0.995168594 |
|
1.00E-05 |
124218.0452 |
102095.9497 |
97192 |
1 |
0.821909164 |
0.782430603 |
1.00E-05 |
123256.9196 |
102725.196 |
97258 |
1 |
0.833423359 |
0.789067261 |
1.007797742 |
0.99387447 |
0.999321393 |
|
9.00E-06 |
136835.3719 |
112952.5578 |
107913 |
1 |
0.825463155 |
0.788633805 |
9.00E-06 |
135849.9447 |
113718.9146 |
108216 |
1 |
0.837092093 |
0.796584792 |
1.007253792 |
0.993260956 |
0.997200044 |
|
8.00E-06 |
152610.0352 |
126633.8995 |
121242 |
1 |
0.829787499 |
0.794456274 |
8.00E-06 |
151453.6382 |
127307.3518 |
121518 |
1 |
0.840569783 |
0.802344542 |
1.00763532 |
0.994710028 |
0.997728732 |
|
7.00E-06 |
172631.2513 |
143969.5829 |
138604 |
1 |
0.833971728 |
0.802890548 |
7.00E-06 |
171409.3065 |
144868.4121 |
139247 |
1 |
0.845160715 |
0.812365459 |
1.007128812 |
0.993795547 |
0.995382306 |
|
6.00E-06 |
199241.1256 |
167092.598 |
161638 |
1 |
0.838645122 |
0.811268254 |
6.00E-06 |
197814.8894 |
168071.0653 |
161784 |
1 |
0.849638093 |
0.817855524 |
1.007209954 |
0.994178253 |
0.999097562 |
|
5.00E-06 |
236247.5427 |
199373.392 |
193006 |
1 |
0.843917315 |
0.816965111 |
5.00E-06 |
234464.0503 |
200668.6884 |
193666 |
1 |
0.855861221 |
0.825994432 |
1.007606677 |
0.9935451 |
0.996592071 |
|
4.00E-06 |
291092.995 |
247769.0754 |
240321 |
1 |
0.851168113 |
0.825581529 |
4.00E-06 |
288965.3467 |
249317.9447 |
241691 |
1 |
0.862795306 |
0.836401329 |
1.007362988 |
0.993787574 |
0.994331605 |
|
3.00E-06 |
381793.9749 |
328359.2915 |
320678 |
1 |
0.860043147 |
0.839924203 |
3.00E-06 |
378816.8543 |
330180.2663 |
322513 |
1 |
0.871609229 |
0.851369194 |
1.007858997 |
0.994484907 |
0.994310307 |
|
2.00E-06 |
560602.8794 |
488816 |
481138 |
1 |
0.871947002 |
0.858251032 |
2.00E-06 |
556550.6985 |
491437.4623 |
482398 |
1 |
0.883005742 |
0.866763803 |
1.007280884 |
0.994665726 |
0.997388049 |
|
1.00E-06 |
1088196.508 |
969244.1558 |
962243 |
1 |
0.890688537 |
0.884254813 |
1.00E-06 |
1080983.186 |
974155.5427 |
964074 |
1 |
0.901175481 |
0.89184921 |
1.006672927 |
0.994958313 |
0.998100768 |
|
9.00E-07 |
1203908.025 |
1075569.915 |
1068340 |
1 |
0.893398742 |
0.88739337 |
9.00E-07 |
1208133.623 |
1080634.106 |
1072751 |
1 |
0.894465716 |
0.887940688 |
0.996502375 |
0.995313686 |
0.995888142 |
|
8.00E-07 |
1348428.719 |
1209469.04 |
1202423 |
1 |
0.89694696 |
0.891721589 |
8.00E-07 |
1353904.367 |
1215191.643 |
1206559 |
1 |
0.897546143 |
0.891170033 |
0.995955661 |
0.995290781 |
0.99657207 |
|
7.00E-07 |
1534027.111 |
1380826.05 |
1375220 |
1 |
0.900131451 |
0.896476985 |
7.00E-07 |
1539773.07 |
1387027.477 |
1381749 |
1 |
0.900799932 |
0.897371845 |
0.996268308 |
0.99552898 |
0.995274829 |
|
6.00E-07 |
1780354.844 |
1610095.312 |
1601415 |
1 |
0.904367642 |
0.899492034 |
6.00E-07 |
1786506.508 |
1616167.709 |
1606431 |
1 |
0.904652573 |
0.899202434 |
0.996556596 |
0.996242719 |
0.99687755 |
|
5.00E-07 |
2125035.533 |
1929071.94 |
1927719 |
1 |
0.907783381 |
0.907146714 |
5.00E-07 |
2133555.362 |
1937818.854 |
1931582 |
1 |
0.90825806 |
0.905334839 |
0.996006746 |
0.995486207 |
0.998000085 |
|
4.00E-07 |
2637184.055 |
2409909.352 |
2410357 |
1 |
0.913819173 |
0.913988918 |
4.00E-07 |
2646545.613 |
2418741.794 |
2414921 |
1 |
0.913924091 |
0.9124804 |
0.996462726 |
0.996348332 |
0.998110083 |
|
3.00E-07 |
3488346.553 |
3209632.161 |
3220585 |
1 |
0.920101289 |
0.923241126 |
3.00E-07 |
3500985.513 |
3222108.256 |
3226161 |
1 |
0.920343213 |
0.921500814 |
0.996389885 |
0.996127971 |
0.99827163 |
|
2.00E-07 |
5184203.387 |
4811423.382 |
4829216 |
1 |
0.928093098 |
0.931525181 |
2.00E-07 |
5199169.422 |
4828569.704 |
4839273 |
1 |
0.928719438 |
0.930778093 |
0.997121457 |
0.996448985 |
0.997921795 |
|
1.00E-07 |
1.02E+07 |
9627046.156 |
9688361 |
1 |
0.943828055 |
0.949839314 |
1.00E-07 |
1.03E+07 |
9653872.432 |
9700878 |
1 |
0.937269168 |
0.941832816 |
0.990291262 |
0.99722119 |
0.998709704 |
很明显当δ<0.1时
r-2-10-2-10-2 > r-3-10-3-10-3 > r-4-10-4-10-4
也就是输入层节点数越少迭代次数越多。
与当输入固定为0.1的数据比较
当δ<0.01时
这组数据同样满足输入层节点数越少迭代次数越多的规律。
再比较随机输入与固定输入迭代次数之间的比例关系（r）-x-10-x-10-x-（x*k），k∈{0,1} / 0.1-x-10-x-10-x-（x*k），k∈{0,1}
从图中明显的观察到随机输入网络（r）-x-10-x-10-x-（x*k），k∈{0,1}的迭代次数是要稍小于固定输入网络0.1-x-10-x-10-x-（x*k），k∈{0,1}的迭代次数。随机输入网络的迭代次数大概要比固定输入0.1的网络的迭代次数要少0.5%。
然后再比较3层网络的数据
|
r-2-10-2 |
r-3-10-3 |
r-4-10-4 |
0.1-2-10-2 |
0.1-3-10-3 |
0.1-4-10-4 |
|||||||||||
|
δ |
迭代次数n |
迭代次数n |
迭代次数n |
r2/r2 |
r3/r2 |
r4/r2 |
δ |
迭代次数n |
迭代次数n |
迭代次数n |
0.1-2/0.1-2 |
0.1-3/0.12 |
0.1-4/0.1-2 |
r-2/0.1-2 |
r-3/0.1-3 |
r-4/0.1-4 |
|
0.5 |
1.869346734 |
2.040201005 |
3 |
1 |
1.091397849 |
1.604838709 |
0.5 |
1.834170854 |
2.150753769 |
4 |
1 |
1.17260274 |
2.180821918 |
1.019178083 |
0.948598131 |
0.75 |
|
0.4 |
5.316582915 |
5.577889447 |
7 |
1 |
1.049149338 |
1.316635161 |
0.4 |
5.467336683 |
5.668341709 |
7 |
1 |
1.036764706 |
1.280330882 |
0.972426471 |
0.984042553 |
1 |
|
0.3 |
10.34673367 |
10.4321608 |
12 |
1 |
1.008256435 |
1.159786304 |
0.3 |
10.3718593 |
10.52763819 |
12 |
1 |
1.015019379 |
1.156976744 |
0.997577519 |
0.990930787 |
1 |
|
0.2 |
18.5678392 |
18.48241206 |
20 |
1 |
0.995399188 |
1.077131258 |
0.2 |
18.98492462 |
19.09547739 |
21 |
1 |
1.005823187 |
1.106140815 |
0.978030704 |
0.967894737 |
0.952380952 |
|
0.1 |
39.7839196 |
38.00502513 |
42 |
1 |
0.955286093 |
1.055702918 |
0.1 |
41.81407035 |
41.92964824 |
43 |
1 |
1.002764091 |
1.028361976 |
0.951448143 |
0.906399808 |
0.976744186 |
|
0.01 |
326.8140704 |
281.0904523 |
347 |
1 |
0.860092872 |
1.061765791 |
0.01 |
405.9145729 |
400.9246231 |
397 |
1 |
0.987706897 |
0.978038303 |
0.805130173 |
0.701105485 |
0.874055416 |
|
0.001 |
2555.768844 |
1994.175879 |
2942 |
1 |
0.78026457 |
1.151121318 |
0.001 |
3909.160804 |
3785.150754 |
3660 |
1 |
0.968277066 |
0.936262329 |
0.653789642 |
0.526841864 |
0.803825137 |
|
1.00E-04 |
21902.38693 |
16359.0402 |
23884 |
1 |
0.74690673 |
1.090474754 |
1.00E-04 |
38025.66332 |
35887.74372 |
33662 |
1 |
0.943776928 |
0.885244255 |
0.575989609 |
0.455839195 |
0.709524092 |
|
9.00E-05 |
23880.13568 |
18649.9598 |
37315 |
1 |
0.780982154 |
1.562595812 |
9.00E-05 |
42192.34171 |
39777.31156 |
37338 |
1 |
0.94276141 |
0.884947327 |
0.565982705 |
0.468859233 |
0.999384006 |
|
8.00E-05 |
27373.11558 |
20427.95477 |
26333 |
1 |
0.74627803 |
0.962002295 |
8.00E-05 |
47395.70352 |
44621.80905 |
41802 |
1 |
0.941473715 |
0.881978679 |
0.57754424 |
0.45780203 |
0.629945936 |
|
7.00E-05 |
30471.34171 |
23575.8593 |
46405 |
1 |
0.77370598 |
1.522906357 |
7.00E-05 |
54052.65327 |
50836.96482 |
47443 |
1 |
0.940508222 |
0.877718246 |
0.563734431 |
0.463754266 |
0.978121114 |
|
6.00E-05 |
33747.86935 |
26453.01508 |
43844 |
1 |
0.783842524 |
1.299163498 |
6.00E-05 |
62950.38191 |
59079.18593 |
55259 |
1 |
0.938504011 |
0.877818344 |
0.536102694 |
0.44775524 |
0.793427315 |
|
5.00E-05 |
41400.74372 |
31641.95477 |
57098 |
1 |
0.764284695 |
1.379153968 |
5.00E-05 |
75340.25628 |
70575.01005 |
65584 |
1 |
0.936750332 |
0.870504074 |
0.549516895 |
0.448345027 |
0.870608685 |
|
4.00E-05 |
50904.86935 |
38461.94975 |
63706 |
1 |
0.755565238 |
1.251471634 |
4.00E-05 |
93917.48744 |
87725.11558 |
81242 |
1 |
0.934065827 |
0.865035918 |
0.542016942 |
0.438437151 |
0.784151055 |
|
3.00E-05 |
68669.09045 |
51233.01508 |
102785 |
1 |
0.746085535 |
1.496816098 |
3.00E-05 |
124764.8342 |
116155.4975 |
107253 |
1 |
0.930995486 |
0.859641266 |
0.550388183 |
0.441072667 |
0.958341492 |
|
2.00E-05 |
98491.42211 |
74906.1206 |
131495 |
1 |
0.760534461 |
1.335090886 |
2.00E-05 |
186176.5779 |
172366.9598 |
158480 |
1 |
0.925825159 |
0.851234896 |
0.529021552 |
0.434573544 |
0.829726148 |
|
1.00E-05 |
193272.4774 |
144249.7286 |
203355 |
1 |
0.746354217 |
1.0521674 |
1.00E-05 |
368710.8643 |
338333.7889 |
308464 |
1 |
0.917612747 |
0.836601331 |
0.524184384 |
0.426353304 |
0.65925035 |
|
9.00E-06 |
206907.4774 |
158850.8543 |
296829 |
1 |
0.767738587 |
1.434597743 |
9.00E-06 |
408984.0804 |
375023.1005 |
341247 |
1 |
0.916962587 |
0.83437722 |
0.505905944 |
0.423576185 |
0.869836218 |
|
8.00E-06 |
237203.201 |
180718.5427 |
363607 |
1 |
0.761872276 |
1.532892467 |
8.00E-06 |
459388.5276 |
420561.8643 |
382394 |
1 |
0.915481861 |
0.832397801 |
0.516345504 |
0.429707394 |
0.950870045 |
|
7.00E-06 |
269616.4372 |
202380.0402 |
421455 |
1 |
0.75062204 |
1.563165081 |
7.00E-06 |
524038.9497 |
478899.1457 |
435268 |
1 |
0.913861739 |
0.830602382 |
0.514496942 |
0.422594281 |
0.968265528 |
|
6.00E-06 |
308451.3015 |
240200 |
405253 |
1 |
0.778729086 |
1.313831383 |
6.00E-06 |
609813.0704 |
556314.598 |
504206 |
1 |
0.912270702 |
0.826820586 |
0.505812874 |
0.431770083 |
0.803744898 |
|
5.00E-06 |
366383.0905 |
279138.8844 |
438198 |
1 |
0.761877094 |
1.196010437 |
5.00E-06 |
729919.3065 |
664199.0553 |
600920 |
1 |
0.909962306 |
0.823269086 |
0.50195013 |
0.420263899 |
0.729211875 |
|
4.00E-06 |
441795.6482 |
344849.6935 |
733902 |
1 |
0.780563808 |
1.661179785 |
4.00E-06 |
909130.6181 |
825530.8241 |
744853 |
1 |
0.908044243 |
0.819302513 |
0.485953987 |
0.417730851 |
0.98529777 |
|
3.00E-06 |
587062.9196 |
459529.0302 |
747879 |
1 |
0.78275942 |
1.273933296 |
3.00E-06 |
1207124.558 |
1091742.894 |
980381 |
1 |
0.904416107 |
0.812162252 |
0.486331684 |
0.420913232 |
0.762845261 |
|
2.00E-06 |
881343.3819 |
672539.7437 |
1245116 |
1 |
0.763084806 |
1.412747886 |
2.00E-06 |
1799195.739 |
1619069.065 |
1449669 |
1 |
0.899884893 |
0.805731677 |
0.489854085 |
0.415386692 |
0.858896755 |
|
1.00E-06 |
1691667.427 |
1292568.874 |
2046819 |
1 |
0.764079779 |
1.209941722 |
1.00E-06 |
3559300.412 |
3175825.116 |
2818734 |
1 |
0.892261048 |
0.791934839 |
0.475280879 |
0.407002535 |
0.726148335 |
|
9.00E-07 |
1861192.327 |
1469246.196 |
2010564 |
1 |
0.789411269 |
1.080255904 |
9.00E-07 |
3947484.774 |
3517441.141 |
3123123 |
1 |
0.891058824 |
0.791167839 |
0.471488159 |
0.417703136 |
0.643767152 |
|
8.00E-07 |
2080246.372 |
1621744.271 |
2166817 |
1 |
0.779592404 |
1.041615565 |
8.00E-07 |
4433661.618 |
3943425.156 |
3493458 |
1 |
0.889428535 |
0.787939699 |
0.469193762 |
0.411252707 |
0.620249907 |
|
7.00E-07 |
2464329.97 |
1797949.739 |
3092601 |
1 |
0.72958969 |
1.254945984 |
7.00E-07 |
5056311.834 |
4490016.492 |
3972648 |
1 |
0.888002291 |
0.785680973 |
0.487376976 |
0.40043277 |
0.778473451 |
|
6.00E-07 |
2925796.92 |
2091886.859 |
2729326 |
1 |
0.714980197 |
0.93284875 |
6.00E-07 |
5882920.477 |
5216202.241 |
4605197 |
1 |
0.886668834 |
0.782807964 |
0.497337493 |
0.401036379 |
0.59266216 |
|
5.00E-07 |
3289760.563 |
2467525.844 |
3698517 |
1 |
0.750062443 |
1.124251121 |
5.00E-07 |
7040336.97 |
6225307.176 |
5492530 |
1 |
0.884234264 |
0.780151578 |
0.467273168 |
0.396370135 |
0.673372198 |
|
4.00E-07 |
4007427.271 |
3216941.045 |
5727957 |
1 |
0.802744711 |
1.429335235 |
4.00E-07 |
8766715.804 |
7731842.281 |
6800941 |
1 |
0.881954252 |
0.775768389 |
0.457118419 |
0.416063976 |
0.842230068 |
|
3.00E-07 |
5245088.296 |
4254122.518 |
6026219 |
1 |
0.811067856 |
1.14892613 |
3.00E-07 |
1.16E+07 |
1.02E+07 |
8973350 |
1 |
0.879310345 |
0.773564655 |
0.452162784 |
0.417070835 |
0.671568478 |
|
2.00E-07 |
7786895.241 |
5986582.417 |
1.07E+07 |
1 |
0.768802229 |
1.374103499 |
2.00E-07 |
1.73E+07 |
1.52E+07 |
1.32E+07 |
1 |
0.878612717 |
0.76300578 |
0.450109552 |
0.393854106 |
0.810606061 |
|
1.00E-07 |
1.58E+07 |
1.23E+07 |
2.71E+07 |
1 |
0.778481013 |
1.715189873 |
1.00E-07 |
3.42E+07 |
2.97E+07 |
2.58E+07 |
1 |
0.868421053 |
0.754385965 |
0.461988304 |
0.414141414 |
1.050387597 |
3层随机输入数据比较
r-2-10-2>r-3-10-3但r-4-10-4的数据变的不规则
3层固定输入的数据
这个明显的0.1-2-10-2>0.1-3-10-3>0.1-4-10-4,输入节点数越少迭代次数越多。
比较随机输入和固定输入的比值
若不考虑r-4和0.1-4的值
r-2/0.1-2 > r-3/0.1-3 的值但都接近0.5.
再比较3层网络和5层网络的迭代次数的比值
|
r-2-10-2 |
r-3-10-3 |
r-4-10-4 |
0.1-2-10-2 |
0.1-3-10-3 |
0.1-4-10-4 |
||
|
δ |
r-2-10-2-10-2 |
r-3-10-3-10-3 |
r-4-10-4-10-4 |
δ |
0.1-2-10-2-10-2 |
0.1-3-10-3-10-3 |
0.1-4-10-4-10-4 |
|
0.5 |
0.981530343 |
1.091397849 |
1 |
0.5 |
0.948051948 |
1.097435898 |
1.333333333 |
|
0.4 |
0.995296331 |
1.046182846 |
1 |
0.4 |
1.01777362 |
1.061147695 |
1 |
|
0.3 |
1 |
1.02014742 |
1 |
0.3 |
1.009784737 |
1.034057256 |
1.090909091 |
|
0.2 |
0.992479184 |
1.009607467 |
1 |
0.2 |
1.017232095 |
1.030648224 |
1.05 |
|
0.1 |
0.988389513 |
0.990310331 |
1.105263158 |
0.1 |
1.04273183 |
1.092861821 |
1.102564103 |
|
0.01 |
1.003100178 |
1.046138022 |
1.476595745 |
0.01 |
1.246616356 |
1.493187603 |
1.675105485 |
|
0.001 |
1.121158815 |
1.186658055 |
2.104434907 |
0.001 |
1.714457303 |
2.247204587 |
2.603129445 |
|
1.00E-04 |
1.376448467 |
1.36274717 |
2.255335222 |
1.00E-04 |
2.401295672 |
2.976551562 |
3.172966349 |
|
9.00E-05 |
1.37053397 |
1.414015331 |
3.196693224 |
9.00E-05 |
2.432948622 |
3.001282685 |
3.186651873 |
|
8.00E-05 |
1.419609413 |
1.393148111 |
2.018937361 |
8.00E-05 |
2.467482608 |
3.027367109 |
3.196604726 |
|
7.00E-05 |
1.405031172 |
1.424993971 |
3.139078671 |
7.00E-05 |
2.506736848 |
3.056305005 |
3.196106171 |
|
6.00E-05 |
1.361443394 |
1.389718279 |
2.559785147 |
6.00E-05 |
2.55099204 |
3.086995622 |
3.213293016 |
|
5.00E-05 |
1.421832869 |
1.406756812 |
2.801943272 |
5.00E-05 |
2.600194485 |
3.12098031 |
3.20234375 |
|
4.00E-05 |
1.436488022 |
1.392058101 |
2.529521541 |
4.00E-05 |
2.661307858 |
3.157454011 |
3.210384889 |
|
3.00E-05 |
1.499024465 |
1.420236783 |
3.095094703 |
3.00E-05 |
2.738129713 |
3.200298169 |
3.219070773 |
|
2.00E-05 |
1.492552806 |
1.419123717 |
2.67108818 |
2.00E-05 |
2.839958089 |
3.244711066 |
3.203687232 |
|
1.00E-05 |
1.555913049 |
1.412883949 |
2.092301836 |
1.00E-05 |
2.991400933 |
3.293581342 |
3.171605421 |
|
9.00E-06 |
1.512090584 |
1.406350218 |
2.750632454 |
9.00E-06 |
3.010557577 |
3.297807597 |
3.153387669 |
|
8.00E-06 |
1.554309326 |
1.42709451 |
2.999018492 |
8.00E-06 |
3.033195723 |
3.303515927 |
3.146809526 |
|
7.00E-06 |
1.561805497 |
1.405713875 |
3.040713111 |
7.00E-06 |
3.057237442 |
3.305752709 |
3.125869857 |
|
6.00E-06 |
1.548130691 |
1.437526275 |
2.507164157 |
6.00E-06 |
3.082746057 |
3.309996263 |
3.1165381 |
|
5.00E-06 |
1.550844027 |
1.400080932 |
2.270385377 |
5.00E-06 |
3.113139543 |
3.309928722 |
3.102867824 |
|
4.00E-06 |
1.517713088 |
1.391818947 |
3.053840488 |
4.00E-06 |
3.146157934 |
3.311156865 |
3.081840035 |
|
3.00E-06 |
1.537643227 |
1.39947016 |
2.332180567 |
3.00E-06 |
3.186565076 |
3.306505583 |
3.03981855 |
|
2.00E-06 |
1.572134954 |
1.375854603 |
2.587856291 |
2.00E-06 |
3.232761622 |
3.294557679 |
3.005130618 |
|
1.00E-06 |
1.55456061 |
1.333584388 |
2.127133167 |
1.00E-06 |
3.29265104 |
3.260080117 |
2.923773486 |
|
9.00E-07 |
1.545958901 |
1.366016449 |
1.881951439 |
9.00E-07 |
3.267423983 |
3.254978833 |
2.911321453 |
|
8.00E-07 |
1.542718827 |
1.34087291 |
1.80204221 |
8.00E-07 |
3.274722887 |
3.245105559 |
2.895389285 |
|
7.00E-07 |
1.606444862 |
1.30208272 |
2.248804555 |
7.00E-07 |
3.283803265 |
3.237150357 |
2.875086575 |
|
6.00E-07 |
1.643378526 |
1.299231694 |
1.704321491 |
6.00E-07 |
3.292974557 |
3.227512969 |
2.866725679 |
|
5.00E-07 |
1.548096732 |
1.279125881 |
1.918597576 |
5.00E-07 |
3.299814523 |
3.212533082 |
2.843539648 |
|
4.00E-07 |
1.519585735 |
1.334880518 |
2.376393621 |
4.00E-07 |
3.312512643 |
3.196638145 |
2.816216762 |
|
3.00E-07 |
1.503602987 |
1.325423695 |
1.871156638 |
3.00E-07 |
3.313352757 |
3.165629206 |
2.781432793 |
|
2.00E-07 |
1.502042775 |
1.244243531 |
2.215680558 |
2.00E-07 |
3.32745456 |
3.147930118 |
2.727682443 |
|
1.00E-07 |
1.549019608 |
1.277650465 |
2.79717075 |
1.00E-07 |
3.32038835 |
3.07648565 |
2.659553084 |
随机输入3层网络的迭代次数是对应5层网络的迭代次数的大约1.5倍左右
固定输入3层网络的迭代次数是对应5层网络的迭代次数的3倍左右。
这组实验表明3层网络随机输入的迭代次数比固定输入的迭代次数要小的多，甚至小于后者的50%。
而5层网络的随机输入的迭代次数只比固定输入的迭代次数略小，二者相差甚至小于1%,也就是5层网络对输入数据并不敏感。
|
实验数据 |
|
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)ti1/1000); |
dr-2-10-2-10-2
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
0.527608315 0.472339838 1.904522613 0 0.5 0.412060302 82 0.001366667
0.619745309 0.379034134 5.341708543 0 0.4 0.391959799 78 0.0013
0.712427458 0.285357852 10.34673367 0 0.3 0.155778894 31 0.000516667
0.806808233 0.19310006 18.70854271 0 0.2 0.236180905 47 0.000783333
0.902011379 0.097776184 40.25125628 0 0.1 0.688442211 142 0.002366667
0.990027433 0.009968527 325.8040201 0 0.01 3.015075377 600 0.01
0.999000529 1.00E-03 2279.577889 0 0.001 15.2361809 3047 0.050783333
0.999900012 1.00E-04 15912.24623 0 1.00E-04 91.9798995 18304 0.305066667
0.999910011 9.00E-05 17423.96482 0 9.00E-05 98.45728643 19609 0.326816667
0.999920009 8.00E-05 19282.14573 0 8.00E-05 109.1306533 21733 0.362216667
0.999930008 7.00E-05 21687.30653 0 7.00E-05 122.4623116 24370 0.406166667
0.999940006 6.00E-05 24788.30151 0 6.00E-05 140.6532663 27990 0.4665
0.999950004 5.00E-05 29117.86935 0 5.00E-05 164.2462312 32700 0.545
0.999960004 4.00E-05 35437.03015 0 4.00E-05 199.3969849 39696 0.6616
0.999970003 3.00E-05 45809.18593 0 3.00E-05 259.8492462 51720 0.862
0.999980002 2.00E-05 65988.56784 0 2.00E-05 375.7236181 74778 1.2463
0.999990001 1.00E-05 124218.0452 0 1.00E-05 703.4974874 140005 2.333416667
0.999991001 9.00E-06 136835.3719 0 9.00E-06 761.959799 151639 2.527316667
0.999992001 8.00E-06 152610.0352 0 8.00E-06 861.0150754 171358 2.855966667
0.999993001 7.00E-06 172631.2513 0 7.00E-06 972.6281407 193560 3.226
0.999994001 6.00E-06 199241.1256 0 6.00E-06 1125.437186 223962 3.7327
0.999995001 5.00E-06 236247.5427 0 5.00E-06 1336.170854 265906 4.431766667
0.999996001 4.00E-06 291092.995 0 4.00E-06 1650.854271 328528 5.475466667
0.999997 3.00E-06 381793.9749 0 3.00E-06 2128.020101 423493 7.058216667
0.999998 2.00E-06 560602.8794 0 2.00E-06 3168.80402 630608 10.51013333
0.999999 1.00E-06 1088196.508 0 1.00E-06 6009.703518 1195947 19.93245
0.9999991 9.00E-07 1203908.025 0 9.00E-07 6628.276382 1319028 21.9838
0.9999992 8.00E-07 1348428.719 0 8.00E-07 7429.336683 1478438 24.64063333
0.9999993 7.00E-07 1534027.111 0 7.00E-07 8488.236181 1689160 28.15266667
0.9999994 6.00E-07 1780354.844 0 6.00E-07 10088.17588 2007549 33.45915
0.9999995 5.00E-07 2125035.533 0 5.00E-07 11910.59799 2370210 39.5035
0.9999996 4.00E-07 2637184.055 0 4.00E-07 14739.1809 2933128 48.88546667
0.9999997 3.00E-07 3488346.553 0 3.00E-07 19689.76382 3918288 65.3048
0.9999998 2.00E-07 5184203.387 0 2.00E-07 29417.75377 5854135 97.56891667
0.9999999 1.00E-07 1.02E+07 0 1.00E-07 54512.9598 10848096 180.8016
607.63275
dr-3-10-3-10-3
f2[0] f2[1] f2[2] 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
0.526440336 0.472256796 0.472675861 1.869346734 0 0.5 0.472361809 109 0.001816667
0.620190597 0.378660228 0.381783874 5.331658291 0 0.4 0.155778894 31 0.000516667
0.713801296 0.285566484 0.287647306 10.22613065 0 0.3 0.236180905 63 0.00105
0.806969107 0.192119984 0.194035873 18.30653266 0 0.2 0.236180905 47 0.000783333
0.902155298 0.097590075 0.097948187 38.37688442 0 0.1 0.472361809 94 0.001566667
0.99004019 0.009950875 0.009949246 268.6934673 0 0.01 2.834170854 564 0.0094
0.99900074 9.99E-04 9.99E-04 1680.497487 0 0.001 13.06030151 2600 0.043333333
0.999900025 1.00E-04 1.00E-04 12004.45729 0 1.00E-04 78.81909548 15700 0.261666667
0.999910028 9.00E-05 9.00E-05 13189.36181 0 9.00E-05 77.22613065 15383 0.256383333
0.999920025 8.00E-05 8.00E-05 14663.1608 0 8.00E-05 85.75376884 17067 0.28445
0.999930019 7.00E-05 7.00E-05 16544.53266 0 7.00E-05 97.30150754 19378 0.322966667
0.999940016 6.00E-05 6.00E-05 19034.80402 0 6.00E-05 113.3366834 22554 0.3759
0.999950014 5.00E-05 5.00E-05 22492.8392 0 5.00E-05 133.7688442 26620 0.443666667
0.999960011 4.00E-05 4.00E-05 27629.55779 0 4.00E-05 163.120603 32476 0.541266667
0.999970009 3.00E-05 3.00E-05 36073.57286 0 3.00E-05 211.8140704 42151 0.702516667
0.999980007 2.00E-05 2.00E-05 52783.36181 0 2.00E-05 309.5376884 61614 1.0269
0.999990003 1.00E-05 1.00E-05 102095.9497 0 1.00E-05 599.5125628 119303 1.988383333
0.999991002 9.00E-06 9.00E-06 112952.5578 0 9.00E-06 670.1407035 133358 2.222633333
0.999992002 8.00E-06 8.00E-06 126633.8995 0 8.00E-06 757.7085427 150784 2.513066667
0.999993002 7.00E-06 7.00E-06 143969.5829 0 7.00E-06 875.2311558 174172 2.902866667
0.999994002 6.00E-06 6.00E-06 167092.598 0 6.00E-06 1003.854271 199767 3.32945
0.999995002 5.00E-06 5.00E-06 199373.392 0 5.00E-06 1183.085427 235434 3.9239
0.999996001 4.00E-06 4.00E-06 247769.0754 0 4.00E-06 1512.356784 300976 5.016266667
0.999997001 3.00E-06 3.00E-06 328359.2915 0 3.00E-06 1945.427136 387140 6.452333333
0.999998001 2.00E-06 2.00E-06 488816 0 2.00E-06 2507.829146 499074 8.3179
0.999999 1.00E-06 1.00E-06 969244.1558 0 1.00E-06 5722.160804 1138710 18.9785
0.9999991 9.00E-07 9.00E-07 1075569.915 0 9.00E-07 6469.055276 1287342 21.4557
0.9999992 8.00E-07 8.00E-07 1209469.04 0 8.00E-07 7172.030151 1427234 23.78723333
0.9999993 7.00E-07 7.00E-07 1380826.05 0 7.00E-07 8186.537688 1629128 27.15213333
0.9999994 6.00E-07 6.00E-07 1610095.312 0 6.00E-07 9600.678392 1910545 31.84241667
0.9999995 5.00E-07 5.00E-07 1929071.94 0 5.00E-07 11807.44221 2349681 39.16135
0.9999996 4.00E-07 4.00E-07 2409909.352 0 4.00E-07 14563.61809 2898167 48.30278333
0.9999997 3.00E-07 3.00E-07 3209632.161 0 3.00E-07 20176.22111 4015080 66.918
0.9999998 2.00E-07 2.00E-07 4811423.382 0 2.00E-07 28304.31156 5632558 93.87596667
0.9999999 1.00E-07 1.00E-07 9627046.156 0 1.00E-07 56264.20603 11196577 186.6096167
599.0246833
dr-4-10-4-10-4
f2[0] f2[1] f2[2] f2[3] 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
0.526261475 0.470138296 0.473433824 0.470196578 3 0 0.5 0.72361809 144 0.0024
0.619481649 0.377210594 0.378727222 0.378469657 7 0 0.4 0.155778894 47 0.000783333
0.713542792 0.285969096 0.285711052 0.286628343 12 0 0.3 0.316582915 63 0.00105
0.808050416 0.191568156 0.192380965 0.19206912 20 0 0.2 0.391959799 78 0.0013
0.902437476 0.097327741 0.097530188 0.097340394 38 0 0.1 0.467336683 109 0.001816667
0.990051834 0.00994703 0.009956631 0.009950273 235 0 0.01 2.462311558 507 0.00845
0.999001067 9.99E-04 9.99E-04 9.99E-04 1398 0 0.001 12.98492462 2584 0.043066667
0.999900049 1.00E-04 9.99E-05 9.99E-05 10590 0 1.00E-04 74.13065327 14752 0.245866667
0.999910044 9.00E-05 9.00E-05 9.00E-05 11673 0 9.00E-05 77.95979899 15529 0.258816667
0.999920041 8.00E-05 8.00E-05 8.00E-05 13043 0 8.00E-05 88.72361809 17657 0.294283333
0.999930029 7.00E-05 7.00E-05 7.00E-05 14783 0 7.00E-05 100.798995 20075 0.334583333
0.999940023 6.00E-05 6.00E-05 6.00E-05 17128 0 6.00E-05 115.5527638 22995 0.38325
0.999950023 5.00E-05 5.00E-05 5.00E-05 20378 0 5.00E-05 141.6733668 28193 0.469883333
0.999960015 4.00E-05 4.00E-05 4.00E-05 25185 0 4.00E-05 175.1155779 34849 0.580816667
0.999970011 3.00E-05 3.00E-05 3.00E-05 33209 0 3.00E-05 230.638191 45897 0.76495
0.999980007 2.00E-05 2.00E-05 2.00E-05 49229 0 2.00E-05 339.8190955 67631 1.127183333
0.999990004 1.00E-05 9.99E-06 1.00E-05 97192 0 1.00E-05 671.6934673 133680 2.228
0.999991003 9.00E-06 9.00E-06 9.00E-06 107913 0 9.00E-06 740.879397 147437 2.457283333
0.999992003 8.00E-06 8.00E-06 8.00E-06 121242 0 8.00E-06 837.6030151 166685 2.778083333
0.999993002 7.00E-06 7.00E-06 7.00E-06 138604 0 7.00E-06 956.9949749 190442 3.174033333
0.999994002 6.00E-06 6.00E-06 6.00E-06 161638 0 6.00E-06 1119.080402 222697 3.711616667
0.999995002 5.00E-06 5.00E-06 5.00E-06 193006 0 5.00E-06 1340.954774 266862 4.4477
0.999996001 4.00E-06 4.00E-06 4.00E-06 240321 0 4.00E-06 1673.728643 333072 5.5512
0.999997001 3.00E-06 3.00E-06 3.00E-06 320678 0 3.00E-06 2218.623116 441506 7.358433333
0.999998 2.00E-06 2.00E-06 2.00E-06 481138 0 2.00E-06 3369.929648 670625 11.17708333
0.999999 1.00E-06 1.00E-06 1.00E-06 962243 0 1.00E-06 6656.326633 1324615 22.07691667
0.9999991 9.00E-07 9.00E-07 9.00E-07 1068340 0 9.00E-07 7402.738693 1473150 24.5525
0.9999992 8.00E-07 8.00E-07 8.00E-07 1202423 0 8.00E-07 8283.81407 1648481 27.47468333
0.9999993 7.00E-07 7.00E-07 7.00E-07 1375220 0 7.00E-07 9110.095477 1812943 30.21571667
0.9999994 6.00E-07 6.00E-07 6.00E-07 1601415 0 6.00E-07 10525.52261 2094580 34.90966667
0.9999995 5.00E-07 5.00E-07 5.00E-07 1927719 0 5.00E-07 12147.62312 2417392 40.28986667
0.9999996 4.00E-07 4.00E-07 4.00E-07 2410357 0 4.00E-07 15609.35678 3106310 51.77183333
0.9999997 3.00E-07 3.00E-07 3.00E-07 3220585 0 3.00E-07 20894.52261 4158010 69.30016667
0.9999998 2.00E-07 2.00E-07 2.00E-07 4829216 0 2.00E-07 31439.65829 6256492 104.2748667
0.9999999 1.00E-07 1.00E-07 1.00E-07 9688361 0 1.00E-07 64211.86432 12778161 212.96935
665.2375
dr-2-10-2
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
0.526584111 0.470703963 1.869346734 0 0.5 0.472361809 110 0.001833333
0.61933022 0.378788514 5.316582915 0 0.4 0.16080402 47 0.000783333
0.714022291 0.285632622 10.34673367 0 0.3 0.08040201 31 0.000516667
0.806512411 0.192506914 18.5678392 0 0.2 0.16080402 47 0.000783333
0.902097214 0.097758283 39.7839196 0 0.1 0.236180905 47 0.000783333
0.99003051 0.009968748 326.8140704 0 0.01 1.487437186 312 0.0052
0.999000471 1.00E-03 2555.768844 0 0.001 9.201005025 1831 0.030516667
0.999900011 1.00E-04 21902.38693 0 1.00E-04 62.85427136 12508 0.208466667
0.999910009 9.00E-05 23880.13568 0 9.00E-05 63.48743719 12634 0.210566667
0.999920007 8.00E-05 27373.11558 0 8.00E-05 73.70854271 14669 0.244483333
0.999930008 7.00E-05 30471.34171 0 7.00E-05 82.66834171 16466 0.274433333
0.999940006 6.00E-05 33747.86935 0 6.00E-05 89.8241206 17875 0.297916667
0.999950004 5.00E-05 41400.74372 0 5.00E-05 111.1809045 22141 0.369016667
0.999960004 4.00E-05 50904.86935 0 4.00E-05 139.040201 27672 0.4612
0.999970002 3.00E-05 68669.09045 0 3.00E-05 184.2763819 36671 0.611183333
0.999980001 2.00E-05 98491.42211 0 2.00E-05 266.6532663 53096 0.884933333
0.999990001 1.00E-05 193272.4774 0 1.00E-05 524.718593 104451 1.74085
0.999991001 9.00E-06 206907.4774 0 9.00E-06 558.1105528 111064 1.851066667
0.999992001 8.00E-06 237203.201 0 8.00E-06 642.5778894 127873 2.131216667
0.999993 7.00E-06 269616.4372 0 7.00E-06 727.0201005 144692 2.411533333
0.999994 6.00E-06 308451.3015 0 6.00E-06 830.2512563 165220 2.753666667
0.999995 5.00E-06 366383.0905 0 5.00E-06 987.0050251 196414 3.273566667
0.999996 4.00E-06 441795.6482 0 4.00E-06 1193.38191 237499 3.958316667
0.999997 3.00E-06 587062.9196 0 3.00E-06 1594.281407 317262 5.2877
0.999998 2.00E-06 881343.3819 0 2.00E-06 2384.386935 474493 7.908216667
0.999999 1.00E-06 1691667.427 0 1.00E-06 4240.211055 843802 14.06336667
0.9999991 9.00E-07 1861192.327 0 9.00E-07 5102.648241 1015444 16.92406667
0.9999992 8.00E-07 2080246.372 0 8.00E-07 5709.417085 1136174 18.93623333
0.9999993 7.00E-07 2464329.97 0 7.00E-07 6725.005025 1338292 22.30486667
0.9999994 6.00E-07 2925796.92 0 6.00E-07 7890.638191 1570237 26.17061667
0.9999995 5.00E-07 3289760.563 0 5.00E-07 8888.251256 1768762 29.47936667
0.9999996 4.00E-07 4007427.271 0 4.00E-07 10831.90955 2155550 35.92583333
0.9999997 3.00E-07 5245088.296 0 3.00E-07 14310.68342 2847826 47.46376667
0.9999998 2.00E-07 7786895.241 0 2.00E-07 21688.41709 4316003 71.93338333
0.9999999 1.00E-07 1.58E+07 0 1.00E-07 42070.00503 8371946 139.5324333
457.6526833
dr-3-10-3
f2[0] f2[1] f2[2] 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
0.531998108 0.4668063 0.467841829 2.040201005 0 0.5 0.391959799 78 0.0013
0.625916993 0.374693698 0.373937912 5.577889447 0 0.4 0.236180905 78 0.0013
0.717575379 0.283496485 0.282197138 10.4321608 0 0.3 0.155778894 31 0.000516667
0.809919228 0.190881722 0.190713318 18.48241206 0 0.2 0.16080402 47 0.000783333
0.902821459 0.097322419 0.09703025 38.00502513 0 0.1 0.236180905 47 0.000783333
0.990045238 0.009953845 0.009955177 281.0904523 0 0.01 1.567839196 312 0.0052
0.999000849 9.99E-04 9.99E-04 1994.175879 0 0.001 8.261306533 1644 0.0274
0.999900027 1.00E-04 1.00E-04 16359.0402 0 1.00E-04 55.13567839 11020 0.183666667
0.999910024 9.00E-05 9.00E-05 18649.9598 0 9.00E-05 56.92462312 11344 0.189066667
0.999920021 8.00E-05 8.00E-05 20427.95477 0 8.00E-05 60.57286432 12069 0.20115
0.999930016 7.00E-05 7.00E-05 23575.8593 0 7.00E-05 70.74874372 14079 0.23465
0.999940017 6.00E-05 6.00E-05 26453.01508 0 6.00E-05 79.9798995 15931 0.265516667
0.999950013 5.00E-05 5.00E-05 31641.95477 0 5.00E-05 96.27638191 19159 0.319316667
0.999960009 4.00E-05 4.00E-05 38461.94975 0 4.00E-05 116.080402 23100 0.385
0.999970008 3.00E-05 3.00E-05 51233.01508 0 3.00E-05 155.4924623 30943 0.515716667
0.999980005 2.00E-05 2.00E-05 74906.1206 0 2.00E-05 226.8291457 45139 0.752316667
0.999990003 1.00E-05 1.00E-05 144249.7286 0 1.00E-05 437.2713568 87032 1.450533333
0.999991002 9.00E-06 9.00E-06 158850.8543 0 9.00E-06 479.3115578 95383 1.589716667
0.999992002 8.00E-06 8.00E-06 180718.5427 0 8.00E-06 545.5276382 108560 1.809333333
0.999993002 7.00E-06 7.00E-06 202380.0402 0 7.00E-06 622.5527638 123894 2.0649
0.999994001 6.00E-06 6.00E-06 240200 0 6.00E-06 722.5929648 143808 2.3968
0.999995001 5.00E-06 5.00E-06 279138.8844 0 5.00E-06 839.4120603 167050 2.784166667
0.999996001 4.00E-06 4.00E-06 344849.6935 0 4.00E-06 1036.477387 206264 3.437733333
0.999997001 3.00E-06 3.00E-06 459529.0302 0 3.00E-06 1381.035176 274831 4.580516667
0.999998 2.00E-06 2.00E-06 672539.7437 0 2.00E-06 2023.020101 402583 6.709716667
0.999999 1.00E-06 1.00E-06 1292568.874 0 1.00E-06 3888.683417 773855 12.89758333
0.9999991 9.00E-07 9.00E-07 1469246.196 0 9.00E-07 4528.236181 901121 15.01868333
0.9999992 8.00E-07 8.00E-07 1621744.271 0 8.00E-07 5122.98995 1019482 16.99136667
0.9999993 7.00E-07 7.00E-07 1797949.739 0 7.00E-07 5819.79397 1158147 19.30245
0.9999994 6.00E-07 6.00E-07 2091886.859 0 6.00E-07 6800.984925 1353401 22.55668333
0.9999995 5.00E-07 5.00E-07 2467525.844 0 5.00E-07 8110.522613 1613995 26.89991667
0.9999996 4.00E-07 4.00E-07 3216941.045 0 4.00E-07 9953.045226 1980658 33.01096667
0.9999997 3.00E-07 3.00E-07 4254122.518 0 3.00E-07 12569.71859 2501374 41.68956667
0.9999998 2.00E-07 2.00E-07 5986582.417 0 2.00E-07 18259.92462 3633725 60.56208333
0.9999999 1.00E-07 1.00E-07 1.23E+07 0 1.00E-07 38868.19095 7734773 128.9128833
407.7492833
dr-4-10-4
f2[0] f2[1] f2[2] f2[3] 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
0.535716861 0.463529881 0.462201609 0.464102577 3 0 0.5 0.547738693 109 0.001816667
0.626382757 0.370585994 0.369941534 0.372025326 7 0 0.4 0.221105528 51 0.00085
0.718509103 0.281085664 0.281285396 0.281171568 12 0 0.3 0.135678392 27 0.00045
0.810595708 0.189874827 0.189742401 0.188944534 20 0 0.2 0.180904523 40 0.000666667
0.903273004 0.096510407 0.096590552 0.096482455 42 0 0.1 0.311557789 62 0.001033333
0.99006448 0.009931742 0.009934172 0.009929011 347 0 0.01 1.768844221 352 0.005866667
0.999001338 9.99E-04 9.99E-04 9.99E-04 2942 0 0.001 8.376884422 1668 0.0278
0.99990005 1.00E-04 9.99E-05 9.99E-05 23884 0 1.00E-04 51.49748744 10254 0.1709
0.999910042 9.00E-05 9.00E-05 9.00E-05 37315 0 9.00E-05 54.46231156 10858 0.180966667
0.999920041 8.00E-05 8.00E-05 8.00E-05 26333 0 8.00E-05 60.04020101 11964 0.1994
0.99993003 7.00E-05 7.00E-05 7.00E-05 46405 0 7.00E-05 67.23115578 13387 0.223116667
0.999940028 6.00E-05 6.00E-05 6.00E-05 43844 0 6.00E-05 80.77386935 16075 0.267916667
0.999950023 5.00E-05 5.00E-05 5.00E-05 57098 0 5.00E-05 94.61809045 18837 0.31395
0.999960017 4.00E-05 4.00E-05 4.00E-05 63706 0 4.00E-05 117.3819095 23359 0.389316667
0.999970013 3.00E-05 3.00E-05 3.00E-05 102785 0 3.00E-05 152.3819095 30332 0.505533333
0.999980009 2.00E-05 2.00E-05 2.00E-05 131495 0 2.00E-05 223.3517588 44456 0.740933333
0.999990004 1.00E-05 1.00E-05 1.00E-05 203355 0 1.00E-05 432.8542714 86146 1.435766667
0.999991003 9.00E-06 9.00E-06 9.00E-06 296829 0 9.00E-06 485.1005025 96535 1.608916667
0.999992003 8.00E-06 8.00E-06 8.00E-06 363607 0 8.00E-06 553.4522613 110137 1.835616667
0.999993002 7.00E-06 7.00E-06 7.00E-06 421455 0 7.00E-06 624.6884422 124313 2.071883333
0.999994002 6.00E-06 6.00E-06 6.00E-06 405253 0 6.00E-06 720.0854271 143305 2.388416667
0.999995002 5.00E-06 5.00E-06 5.00E-06 438198 0 5.00E-06 846.7085427 168511 2.808516667
0.999996001 4.00E-06 4.00E-06 4.00E-06 733902 0 4.00E-06 1062.241206 211386 3.5231
0.999997001 3.00E-06 3.00E-06 3.00E-06 747879 0 3.00E-06 1375.070352 273639 4.56065
0.999998001 2.00E-06 2.00E-06 2.00E-06 1245116 0 2.00E-06 2037.321608 405442 6.757366667
0.999999 1.00E-06 1.00E-06 1.00E-06 2046819 0 1.00E-06 3900.668342 776234 12.93723333
0.9999991 9.00E-07 9.00E-07 9.00E-07 2010564 0 9.00E-07 4260.743719 847889 14.13148333
0.9999992 8.00E-07 8.00E-07 8.00E-07 2166817 0 8.00E-07 4823.21608 959820 15.997
0.9999993 7.00E-07 7.00E-07 7.00E-07 3092601 0 7.00E-07 5536.517588 1101799 18.36331667
0.9999994 6.00E-07 6.00E-07 6.00E-07 2729326 0 6.00E-07 6414.592965 1276520 21.27533333
0.9999995 5.00E-07 5.00E-07 5.00E-07 3698517 0 5.00E-07 7769.160804 1546067 25.76778333
0.9999996 4.00E-07 4.00E-07 4.00E-07 5727957 0 4.00E-07 9313.638191 1853418 30.8903
0.9999997 3.00E-07 3.00E-07 3.00E-07 6026219 0 3.00E-07 12582.55276 2503943 41.73238333
0.9999998 2.00E-07 2.00E-07 2.00E-07 1.07E+07 0 2.00E-07 18166.89447 3615217 60.25361667
0.9999999 1.00E-07 1.00E-07 1.00E-07 2.71E+07 0 1.00E-07 36836.72864 7330514 122.1752333
393.5444333
本次实验原始数据比较多有感兴趣的朋友可以在我的资源里下载
输入对5层网络迭代次数的影响相关推荐
- 权重初始化方式对神经网络迭代次数的影响
做一个网络 输入X范围是[1e-16,37] 权重初始化方式 Random rand1 =new Random(); int ti1=rand1.nextInt(98)+1; tw[a][b]=(do ...
- 空间尺寸对迭代次数的影响
( A, B )---3*30*2---( 1, 0 )( 0, 1 ) ( A, B )---4*30*2---( 1, 0 )( 0, 1 ) 做4个训练集尺寸分别为3行3列,3行4列,4行3列和 ...
- 神经网络的输入对迭代次数的影响
如果一个神经网络只有一个输入值,当这个输入值大小发生变化的时候对网络的收敛的迭代次数是否有影响. 比如如上的网络输入的x值从1e-16到45 权重的初始化方式是 Random rand1 =new R ...
- 神经网络输出数量对迭代次数的影响
制作一个神经网络输入minst数据集的2的前200张图,经过3*3的卷积核向001收敛 收敛标准是 while(Math.abs(y[0]-0)> δ || Math.abs(y[1]-0)& ...
- 学习对象对神经网络迭代次数的影响
制作一个81*30*2的神经网络 输入minst数据集的1的图片,让这个网络向1,0收敛, y[0]=1,y[1]=0 收敛标准 while (Math.abs(f2[0] - y[0]) > ...
- 数值分布的分散程度对迭代次数的影响
( A, B )---1*30*2---( 1, 0 )( 0, 1 ) 让网络的输入只有1个节点,AB各由7张二值化的图片组成,排列组合A和B的所有可能性,固定收敛误差为7e-4,统计收敛迭代次数 ...
- 学习率对神经网络迭代次数的影响
调整学习率看看对网络的迭代次数是否有影响 学习率先后实验了5, 1, 0.5, 0.3, 0.2, 0.1, 0.01, 0.005, 0.003,0.001 权重的初始化标准是 Random ran ...
- 等位点数值差对迭代次数的影响
移位距离和假设 (A,B)---m*n*k---(1,0)(0,1) 用神经网络分类A和B,把参与分类的A和B中的数字看作是组成A和B的粒子,分类的过程就是让A和B中的粒子互相交换位置,寻找最短移位路 ...
- 测量一组5层网络的迭代次数
如图左边5层网络很显然可以看作是右边的3层网络两个组合而成的,所以左边的5层网络的迭代次数和右边的3层网络的迭代次数有没有什么关系? 5层 3层 2*10*2*10*2 2*10*2 3*10*3*1 ...
最新文章
- java jdk 1.8 安装_下载、安装、配置 java jdk1.8
- Javascript函数之深入浅出递归思想，附案例与代码！
- 一个小灯泡引发大论战：千万粉丝科普up主翻车，伊朗“唐马儒”、李永乐等下场，30万公里导线引百万网友围观...
- 【多线程】阻塞队列的C++多线程 实现 BlockingQueue
- LIVE555再学习 -- VLC搭建RTSP服务器（转）
- Android之基于AssetManager实现换肤方案
- 通用无线设备对码软件_珞光全新发布国产通用软件无线电平台 ：USRP-LW N310！珞光品牌已实现国产替代...
- 六本JAVA架构书，构建科学得架构知识体系
- dubbo服务提供与消费
- app软件测试是否强制升级_这些测试方法对于任何软件都必须是强制性的
- BZOJ2005 [NOI2010]能量采集
- mysql安装步骤以及问题---解压版本
- 修改本机域名localhost为任意你想要的名称
- 群贤路附近哪有计算机学校,袍江,镜湖,皋埠,富盛…绍兴各镇学校2017年施教区划定!你家孩子该在哪就读?...
- html5与java开发视频教程_JAVA视频—WEB开发视频教程
- Android开发应用apk文件发送到微信打不开解决方案
- linux无法访问移动硬盘,移动硬盘“无法访问”的解决方案
- Android 炫酷自定义 View - 剑气加载
- 如何调用WxJump的API接口实现微信网址链接转二维码
- 端口映射和端口转发区别是什么