ANN神经网络分类2*2矩阵：吸引子和反鞍点cfa-cp

最新推荐文章于 2022-11-21 20:21:34 发布

黑榆

最新推荐文章于 2022-11-21 20:21:34 发布

阅读量1.5k

点赞数

分类专栏：用神经网络模拟原子文章标签： ANN 吸引子反鞍点 2*2矩阵神经网络

本文链接：https://blog.csdn.net/georgesale/article/details/90699513

版权

用神经网络模拟原子专栏收录该内容

358 篇文章 5 订阅

订阅专栏

用吸引子和反鞍点做训练集用排斥子和吸引子做测试集，通过改变测试集吸引子和排斥子的比例观察分类效果，并总结规律。

r1	r2
<1	<1	吸引子	c
>1	>1	排斥子	p
>1	<1	鞍点	a
<1	>1	反鞍点	fa

训练集的比例c：fa=1：1，

测试集c和p的比例为x：y.

让x：y的比例分别为0：10，1：9，2：8，3：7，4：6，5：5，6：4，7：3，8：2，9：1，10：0。

训练集不含有p这实际上也同时实现了参数迁移。

实验过程

二分类c和fa

制作一个4*4*2的网络向这个的左侧输入吸引子，并让左侧网络向1，0收敛；向右侧网络输入反鞍点a让右侧向0，1收敛，并让4*4*2部分权重共享，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。

一个数学计算的实例

《神经网络的可能原理---还原论的振子力学系统》

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 )

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。

激活函数用sigmoid，反向传导的差值的计算方法就是Math.abs(f2[n]-y[n])。

本文尝试了δ从0.5到1e-6在内的26个值.

具体进样顺序
进样顺序	迭代次数
δ=0.5
C	1	判断是否达到收敛
Fa	2	判断是否达到收敛
梯度下降
C	3	判断是否达到收敛
Fa	4	判断是否达到收敛
梯度下降
……
达到收敛标准测量准确率,记录迭代次数，将这个过程重复199次
δ=0.4
…
δ=1e-6

将这个网络简写成

d2（c，fa）-4-4-2-（2*k）,k∈{0,1}

吸引子的初始化方式

Random rand1 =new Random();

int ti1=rand1.nextInt(99)+1;

x[0]=sigmoid((double)ti1/100) ;

Random rand2 =new Random();

int ti2=rand2.nextInt(99)+1;

x[3]=sigmoid ((double)ti2/100) ;

反鞍点的初始化方式

Random rand1 =new Random();

int ti1=rand1.nextInt(99)+1;

x[0]=sigmoid ((double)ti1/100) ;

Random rand2 =new Random();

int ti2=rand2.nextInt(99)+1;

x[3]=sigmoid (1+ ((double)ti2/100) );

得到的数据

以第一组为例

训练集	0.5	c
	0.5	fa
测试集	0	c
	1	p	cfacp01
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	最大准确率p-max
0.499607	0.516459	182.6533	1	0.5	1.964824	391	0.006517	1
0.602002	0.402953	4251.07	0.534947	0.4	9.979899	2002	0.033367	0.813814
0.716424	0.287718	4974.915	0.302725	0.3	9.422111	1876	0.031267	0.576577
0.820372	0.183737	5630.246	0.175165	0.2	10.64824	2119	0.035317	0.409409
0.913189	0.088254	6445.503	0.123365	0.1	12.07538	2404	0.040067	0.361361
0.991841	0.008424	9132.859	0.067872	0.01	16.30653	3277	0.054617	0.258258
0.999189	8.28E-04	13198.52	0.058667	0.001	23.39196	4658	0.077633	0.192192
0.999917	8.38E-05	21881.98	0.066021	1.00E-04	37.43719	7450	0.124167	0.17017
0.999926	7.51E-05	22943.3	0.066539	9.00E-05	39.11558	7784	0.129733	0.204204
0.999934	6.65E-05	23115.73	0.062459	8.00E-05	39.57789	7891	0.131517	0.169169
0.999943	5.79E-05	24156.79	0.064788	7.00E-05	41.26131	8211	0.13685	0.204204
0.999951	4.92E-05	25025.05	0.068013	6.00E-05	42.55779	8469	0.14115	0.277277
0.99996	4.04E-05	25553.76	0.06589	5.00E-05	43.43216	8645	0.144083	0.169169
0.999968	3.26E-05	27567.8	0.066821	4.00E-05	46.9397	9357	0.15595	0.181181
0.999976	2.47E-05	28915.17	0.068994	3.00E-05	48.51256	9654	0.1609	0.23023
0.999984	1.64E-05	32992.38	0.067334	2.00E-05	55.48241	11043	0.18405	0.195195
0.999992	8.26E-06	40302.97	0.074492	1.00E-05	67.94472	13522	0.225367	0.182182
0.999993	7.23E-06	42394.87	0.066966	9.00E-06	70.96482	14139	0.23565	0.2002
0.999994	6.39E-06	43147.64	0.06342	8.00E-06	72.29146	14386	0.239767	0.216216
0.999995	5.48E-06	45334.41	0.065538	7.00E-06	75.98995	15122	0.252033	0.224224
0.999995	4.79E-06	48249.27	0.064663	6.00E-06	81.09045	16137	0.26895	0.208208
0.999996	4.12E-06	52311.14	0.064351	5.00E-06	87.41709	17427	0.29045	0.23023
0.999997	3.27E-06	58895.01	0.059175	4.00E-06	99.22111	19762	0.329367	0.217217
0.999998	2.40E-06	67402.08	0.057922	3.00E-06	112.4221	22390	0.373167	0.219219
0.999998	1.57E-06	87757.98	0.062962	2.00E-06	146.6231	29183	0.486383	0.269269
0.999999	7.56E-07	136105.9	0.062897	1.00E-06	226.2261	45020	0.750333	0.322322

这个网络的测试集c与p的比例是0：1，也就全是p。

平均准确率p-ave，是199次收敛的平均值

最大准确率p-max，是199次收敛的最大值

将平均准确率的后10个平均0.064239，这个网络的平均准确率趋向一个常数。

其他的值

平均准确率p-ave
δ
p	1	0.9	0.8	0.7	0.6	0.5	0.4	0.3	0.2	0.1	0
c	0	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9	1
0.5	1	0.899472	0.799473	0.699413	0.60052	0.500335	0.399304	0.301503	0.200487	0.107218	2.06E-04
0.4	0.534947	0.567004	0.589187	0.628463	0.65807	0.687175	0.710308	0.749589	0.776591	0.80816	0.83479
0.3	0.302725	0.351216	0.397931	0.465053	0.502533	0.572819	0.630102	0.686762	0.740877	0.798079	0.85819
0.2	0.175165	0.252172	0.318962	0.389274	0.454636	0.522402	0.586204	0.651853	0.723497	0.783085	0.854991
0.1	0.123365	0.197836	0.269526	0.341522	0.418298	0.487281	0.566174	0.636179	0.707728	0.7849	0.854392
0.01	0.067872	0.149562	0.233198	0.309038	0.388378	0.466602	0.549112	0.627089	0.705414	0.783271	0.866017
0.001	0.058667	0.141106	0.223636	0.30607	0.388016	0.469022	0.552623	0.63392	0.715238	0.800645	0.878542
1.00E-04	0.066021	0.147333	0.229843	0.313374	0.396145	0.479303	0.563232	0.644962	0.727642	0.808492	0.890529
9.00E-05	0.066539	0.145271	0.229229	0.313057	0.399777	0.481295	0.562678	0.645007	0.727396	0.812164	0.894739
8.00E-05	0.062459	0.151911	0.228324	0.313831	0.400989	0.480943	0.566969	0.644252	0.729559	0.810901	0.894226
7.00E-05	0.064788	0.15422	0.229471	0.315084	0.399062	0.482115	0.565661	0.644418	0.729383	0.812687	0.896208
6.00E-05	0.068013	0.14934	0.23347	0.316774	0.39617	0.482191	0.565681	0.646129	0.732698	0.810187	0.896318
5.00E-05	0.06589	0.148953	0.234642	0.317503	0.397257	0.484072	0.565098	0.649212	0.732245	0.815791	0.896218
4.00E-05	0.066821	0.152404	0.233671	0.31782	0.400627	0.483549	0.568926	0.649831	0.734378	0.8156	0.900312
3.00E-05	0.068994	0.148209	0.234496	0.322192	0.402091	0.485491	0.566521	0.651425	0.735982	0.819845	0.901469
2.00E-05	0.067334	0.14927	0.23514	0.318977	0.405285	0.488066	0.571672	0.653895	0.73979	0.82567	0.906072
1.00E-05	0.074492	0.15005	0.238681	0.323193	0.41031	0.492367	0.577075	0.664524	0.745982	0.833623	0.919759
9.00E-06	0.066966	0.150558	0.233555	0.328233	0.407629	0.492332	0.579378	0.662728	0.74892	0.836263	0.920237
8.00E-06	0.06342	0.148073	0.236543	0.323484	0.411326	0.491688	0.581516	0.666405	0.751007	0.836605	0.922209
7.00E-06	0.065538	0.153264	0.236442	0.321935	0.409002	0.492125	0.58215	0.667014	0.752858	0.838034	0.922661
6.00E-06	0.064663	0.153556	0.238057	0.32783	0.411532	0.498227	0.582422	0.66804	0.754247	0.839563	0.926434
5.00E-06	0.064351	0.154104	0.238887	0.320798	0.412775	0.499173	0.584962	0.672839	0.757863	0.846349	0.933813
4.00E-06	0.059175	0.146644	0.236483	0.326316	0.415104	0.501099	0.589821	0.67479	0.765534	0.852863	0.937978
3.00E-06	0.057922	0.148626	0.23943	0.324023	0.416628	0.506024	0.594796	0.685741	0.772748	0.863643	0.95254
2.00E-06	0.062962	0.148797	0.2386	0.3302	0.418247	0.51373	0.604066	0.694011	0.784362	0.87739	0.965639
1.00E-06	0.062897	0.154964	0.245778	0.339344	0.436321	0.526924	0.614982	0.710645	0.80073	0.89415	0.984256

平均值	0.064239	0.150863	0.238246	0.326536	0.414887	0.501369	0.589117	0.676674	0.763425	0.851848	0.938553

平均准确率变化是很有规律的

将平均值画成图是一条直线

很容易求得这条线的表达式

与实测值比较

pp		1	0.9	0.8	0.7	0.6	0.5	0.4	0.3	0.2	0.1	0
pc	δ	0	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9	1
实测值		0.064239	0.150863	0.238246	0.326536	0.414887	0.501369	0.589117	0.676674	0.763425	0.851848	0.938553
计算值0.06423861pp+0.938552623pc		0.064239	0.15167	0.239101	0.326533	0.413964	0.501396	0.588827	0.676258	0.76369	0.851121	0.938553

将实测值和计算值画成图

图像是高度重合的。

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;

tw[a][b]=xx*((double)ti1/100);

训练集	0.5	c
	0.5	fa
测试集	0	c
	1	p	cfacp01
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	最大准确率p-max
0.499607	0.516459	182.6533	1	0.5	1.964824	391	0.006517	1
0.602002	0.402953	4251.07	0.534947	0.4	9.979899	2002	0.033367	0.813814
0.716424	0.287718	4974.915	0.302725	0.3	9.422111	1876	0.031267	0.576577
0.820372	0.183737	5630.246	0.175165	0.2	10.64824	2119	0.035317	0.409409
0.913189	0.088254	6445.503	0.123365	0.1	12.07538	2404	0.040067	0.361361
0.991841	0.008424	9132.859	0.067872	0.01	16.30653	3277	0.054617	0.258258
0.999189	8.28E-04	13198.52	0.058667	0.001	23.39196	4658	0.077633	0.192192
0.999917	8.38E-05	21881.98	0.066021	1.00E-04	37.43719	7450	0.124167	0.17017
0.999926	7.51E-05	22943.3	0.066539	9.00E-05	39.11558	7784	0.129733	0.204204
0.999934	6.65E-05	23115.73	0.062459	8.00E-05	39.57789	7891	0.131517	0.169169
0.999943	5.79E-05	24156.79	0.064788	7.00E-05	41.26131	8211	0.13685	0.204204
0.999951	4.92E-05	25025.05	0.068013	6.00E-05	42.55779	8469	0.14115	0.277277
0.99996	4.04E-05	25553.76	0.06589	5.00E-05	43.43216	8645	0.144083	0.169169
0.999968	3.26E-05	27567.8	0.066821	4.00E-05	46.9397	9357	0.15595	0.181181
0.999976	2.47E-05	28915.17	0.068994	3.00E-05	48.51256	9654	0.1609	0.23023
0.999984	1.64E-05	32992.38	0.067334	2.00E-05	55.48241	11043	0.18405	0.195195
0.999992	8.26E-06	40302.97	0.074492	1.00E-05	67.94472	13522	0.225367	0.182182
0.999993	7.23E-06	42394.87	0.066966	9.00E-06	70.96482	14139	0.23565	0.2002
0.999994	6.39E-06	43147.64	0.06342	8.00E-06	72.29146	14386	0.239767	0.216216
0.999995	5.48E-06	45334.41	0.065538	7.00E-06	75.98995	15122	0.252033	0.224224
0.999995	4.79E-06	48249.27	0.064663	6.00E-06	81.09045	16137	0.26895	0.208208
0.999996	4.12E-06	52311.14	0.064351	5.00E-06	87.41709	17427	0.29045	0.23023
0.999997	3.27E-06	58895.01	0.059175	4.00E-06	99.22111	19762	0.329367	0.217217
0.999998	2.40E-06	67402.08	0.057922	3.00E-06	112.4221	22390	0.373167	0.219219
0.999998	1.57E-06	87757.98	0.062962	2.00E-06	146.6231	29183	0.486383	0.269269
0.999999	7.56E-07	136105.9	0.062897	1.00E-06	226.2261	45020	0.750333	0.322322

训练集	0.5	c
	0.5	fa
测试集	0.1	c
	0.9	p	cfacp02
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	最大准确率p-max
0.499544	0.516503	181.6583	0.899472	0.5	2.125628	454	0.007567	0.940941
0.590557	0.413901	4290.206	0.567004	0.4	10.27136	2060	0.034333	0.816817
0.719535	0.285092	4925.251	0.351216	0.3	9.59799	1910	0.031833	0.611612
0.821236	0.183434	5555.261	0.252172	0.2	10.50251	2091	0.03485	0.438438
0.913695	0.088914	6381.06	0.197836	0.1	11.67337	2323	0.038717	0.378378
0.991744	0.008538	9211.442	0.149562	0.01	16.60302	3304	0.055067	0.288288
0.999187	8.28E-04	13499.47	0.141106	0.001	23.42211	4663	0.077717	0.256256
0.999918	8.30E-05	21984.57	0.147333	1.00E-04	37.73869	7526	0.125433	0.262262
0.999927	7.35E-05	22402.69	0.145271	9.00E-05	37.8995	7543	0.125717	0.243243
0.999935	6.61E-05	23222.02	0.151911	8.00E-05	39.49246	7859	0.130983	0.256256
0.999942	5.83E-05	23942.22	0.15422	7.00E-05	40.69347	8099	0.134983	0.252252
0.999951	4.99E-05	24585.32	0.14934	6.00E-05	41.75879	8312	0.138533	0.262262
0.99996	4.03E-05	26133.32	0.148953	5.00E-05	44.10553	8777	0.146283	0.266266
0.999967	3.30E-05	27834.47	0.152404	4.00E-05	47.50251	9453	0.15755	0.275275
0.999976	2.44E-05	29697.13	0.148209	3.00E-05	49.9196	9934	0.165567	0.247247
0.999984	1.61E-05	33122.31	0.14927	2.00E-05	56.14573	11175	0.18625	0.283283
0.999992	8.22E-06	40501.41	0.15005	1.00E-05	67.64322	13476	0.2246	0.269269
0.999993	7.41E-06	41123.25	0.150558	9.00E-06	68.92965	13717	0.228617	0.254254
0.999994	6.43E-06	43245.48	0.148073	8.00E-06	72.8392	14511	0.24185	0.251251
0.999994	5.80E-06	45612.29	0.153264	7.00E-06	76.18593	15182	0.253033	0.307307
0.999995	4.85E-06	47714.09	0.153556	6.00E-06	82.30653	16381	0.273017	0.273273
0.999996	4.03E-06	51780.28	0.154104	5.00E-06	85.64824	17044	0.284067	0.27027
0.999997	3.19E-06	57009.79	0.146644	4.00E-06	94.25126	18757	0.312617	0.301301
0.999998	2.36E-06	67125.24	0.148626	3.00E-06	110.8643	22062	0.3677	0.268268
0.999998	1.60E-06	88700.8	0.148797	2.00E-06	145.9447	29059	0.484317	0.341341
0.999999	7.37E-07	137744.8	0.154964	1.00E-06	226.206	45030	0.7505	0.345345

训练集	0.5	c
	0.5	fa
测试集	0.2	c
	0.8	p	cfacp03
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	最大准确率p-max
0.499518	0.516502	178.5829	0.799473	0.5	2.19598	437	0.007283	0.826827
0.598625	0.407394	4254.588	0.589187	0.4	8.944724	1780	0.029667	0.77978
0.717806	0.28832	4935.819	0.397931	0.3	9.321608	1855	0.030917	0.620621
0.820199	0.185541	5589.281	0.318962	0.2	10.45729	2081	0.034683	0.524525
0.914598	0.088518	6386.216	0.269526	0.1	11.78894	2346	0.0391	0.443443
0.991723	0.008491	9068.95	0.233198	0.01	18.44724	3687	0.06145	0.332332
0.999193	8.19E-04	13523.46	0.223636	0.001	23.56784	4690	0.078167	0.318318
0.99992	8.10E-05	21775.75	0.229843	1.00E-04	36.66834	7313	0.121883	0.322322
0.999927	7.34E-05	22805.15	0.229229	9.00E-05	38.26633	7615	0.126917	0.329329
0.999934	6.61E-05	23236.22	0.228324	8.00E-05	39.43719	7848	0.1308	0.312312
0.999943	5.82E-05	23843.02	0.229471	7.00E-05	40.09548	7995	0.13325	0.307307
0.999951	4.96E-05	24459.93	0.23347	6.00E-05	40.47236	8054	0.134233	0.332332
0.999959	4.10E-05	25862.86	0.234642	5.00E-05	42.8392	8525	0.142083	0.36036
0.999968	3.25E-05	27466.58	0.233671	4.00E-05	45.41709	9038	0.150633	0.317317
0.999976	2.43E-05	29748.48	0.234496	3.00E-05	49.47739	9846	0.1641	0.32032
0.999984	1.62E-05	32526.07	0.23514	2.00E-05	54.54271	10869	0.18115	0.325325
0.999992	8.23E-06	40372.25	0.238681	1.00E-05	66.8392	13302	0.2217	0.343343
0.999993	7.20E-06	42045.49	0.233555	9.00E-06	69.70854	13872	0.2312	0.329329
0.999994	6.31E-06	43259.52	0.236543	8.00E-06	72.18593	14365	0.239417	0.318318
0.999994	5.70E-06	44939.51	0.236442	7.00E-06	74.19095	14764	0.246067	0.32032
0.999995	4.83E-06	48766.47	0.238057	6.00E-06	81.17085	16153	0.269217	0.349349
0.999996	3.93E-06	50475.07	0.238887	5.00E-06	83.27136	16603	0.276717	0.344344
0.999997	3.20E-06	57375.83	0.236483	4.00E-06	94.9598	18897	0.31495	0.354354
0.999998	2.37E-06	67287.89	0.23943	3.00E-06	111.1709	22123	0.368717	0.361361
0.999998	1.57E-06	87975.37	0.2386	2.00E-06	144.1407	28684	0.478067	0.414414
0.999999	7.57E-07	134449.8	0.245778	1.00E-06	221.5678	44108	0.735133	0.376376

训练集	0.5	c
	0.5	fa
测试集	0.3	c
	0.7	p	cfacp04
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	最大准确率p-max
0.499564	0.516334	179.5678	0.699413	0.5	2.256281	464	0.007733	0.835836
0.598711	0.406128	4203.678	0.628463	0.4	9.125628	1832	0.030533	0.795796
0.719764	0.287777	4996.291	0.465053	0.3	9.286432	1848	0.0308	0.653654
0.819072	0.186145	5596.608	0.389274	0.2	10.30653	2051	0.034183	0.546547
0.913286	0.090043	6386.457	0.341522	0.1	11.57789	2304	0.0384	0.466466
0.991718	0.008582	9132.085	0.309038	0.01	16.48241	3280	0.054667	0.418418
0.999195	8.21E-04	13376.19	0.30607	0.001	22.8191	4541	0.075683	0.379379
0.999918	8.29E-05	21822.06	0.313374	1.00E-04	36.27638	7219	0.120317	0.393393
0.999927	7.40E-05	22558.95	0.313057	9.00E-05	37.9196	7547	0.125783	0.379379
0.999935	6.61E-05	23350.17	0.313831	8.00E-05	38.75879	7729	0.128817	0.39039
0.999944	5.69E-05	23957.44	0.315084	7.00E-05	39.9799	7957	0.132617	0.392392
0.999951	4.88E-05	24828.13	0.316774	6.00E-05	41.1206	8183	0.136383	0.393393
0.999959	4.16E-05	25927.43	0.317503	5.00E-05	43.43719	8644	0.144067	0.383383
0.999967	3.32E-05	27510.07	0.31782	4.00E-05	45.75879	9106	0.151767	0.42042
0.999975	2.48E-05	29454.89	0.322192	3.00E-05	49.0201	9755	0.162583	0.416416
0.999984	1.65E-05	32773.93	0.318977	2.00E-05	54.16583	10779	0.17965	0.4004
0.999992	8.18E-06	39950.8	0.323193	1.00E-05	66.45729	13240	0.220667	0.41041
0.999993	7.53E-06	41977.64	0.328233	9.00E-06	69.87437	13905	0.23175	0.413413
0.999994	6.44E-06	43941.27	0.323484	8.00E-06	72.99497	14526	0.2421	0.405405
0.999994	5.59E-06	45721.48	0.321935	7.00E-06	75.00503	14941	0.249017	0.399399
0.999995	4.83E-06	47848.13	0.32783	6.00E-06	78.8191	15685	0.261417	0.439439
0.999996	4.01E-06	52577.88	0.320798	5.00E-06	86.94975	17304	0.2884	0.428428
0.999997	3.13E-06	57028.19	0.326316	4.00E-06	93.38191	18583	0.309717	0.423423
0.999998	2.40E-06	67301.64	0.324023	3.00E-06	110.809	22067	0.367783	0.385385
0.999998	1.59E-06	87602.23	0.3302	2.00E-06	144.809	28817	0.480283	0.439439
0.999999	7.51E-07	130951	0.339344	1.00E-06	215.804	42945	0.71575	0.535536

训练集	0.5	c
	0.5	fa
测试集	0.4	c
	0.6	p	cfacp05
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	最大准确率p-max
0.499545	0.516043	179.6683	0.60052	0.5	1.879397	391	0.006517	0.772773
0.59702	0.407705	4164.869	0.65807	0.4	9.462312	1883	0.031383	0.822823
0.718879	0.285339	4995.171	0.502533	0.3	9.477387	1888	0.031467	0.662663
0.819866	0.185018	5522.156	0.454636	0.2	10.22111	2034	0.0339	0.605606
0.913223	0.088643	6483.412	0.418298	0.1	11.65327	2319	0.03865	0.504505
0.991752	0.008458	9001.392	0.388378	0.01	16.56784	3297	0.05495	0.466466
0.999165	8.43E-04	13524.71	0.388016	0.001	23.0402	4602	0.0767	0.451451
0.999918	8.31E-05	21694.95	0.396145	1.00E-04	36.47739	7259	0.120983	0.448448
0.999926	7.44E-05	22502.05	0.399777	9.00E-05	37.61307	7501	0.125017	0.476476
0.999935	6.56E-05	23194.56	0.400989	8.00E-05	39.1407	7789	0.129817	0.484484
0.999942	5.86E-05	23595.37	0.399062	7.00E-05	39.56784	7874	0.131233	0.454454
0.999951	4.92E-05	24682.91	0.39617	6.00E-05	41.37186	8233	0.137217	0.462462
0.99996	4.00E-05	26237.58	0.397257	5.00E-05	44.63819	8883	0.14805	0.446446
0.999968	3.22E-05	27352.37	0.400627	4.00E-05	46.06533	9167	0.152783	0.482482
0.999975	2.50E-05	29533.8	0.402091	3.00E-05	49.42211	9867	0.16445	0.47047
0.999984	1.64E-05	32713.57	0.405285	2.00E-05	54.98995	10943	0.182383	0.47047
0.999992	8.11E-06	40253.63	0.41031	1.00E-05	67.51759	13436	0.223933	0.486486
0.999993	7.44E-06	41666.47	0.407629	9.00E-06	69.8191	13909	0.231817	0.473473
0.999994	6.32E-06	42947.6	0.411326	8.00E-06	71.77387	14283	0.23805	0.501502
0.999994	5.52E-06	46115.12	0.409002	7.00E-06	77.06533	15336	0.2556	0.471471
0.999995	4.84E-06	47683.86	0.411532	6.00E-06	79.04523	15730	0.262167	0.475475
0.999996	3.99E-06	51373.28	0.412775	5.00E-06	85.49246	17013	0.28355	0.468468
0.999997	3.18E-06	56097.63	0.415104	4.00E-06	92.8191	18471	0.30785	0.496496
0.999998	2.41E-06	67657.76	0.416628	3.00E-06	112.0452	22297	0.371617	0.498498
0.999998	1.56E-06	90365.29	0.418247	2.00E-06	149.8794	29841	0.49735	0.499499
0.999999	7.35E-07	136429.8	0.436321	1.00E-06	224.397	44670	0.7445	0.585586

训练集	0.5	c
	0.5	fa
测试集	0.5	c
	0.5	p	cfacp06
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	最大准确率p-max
0.49958	0.515852	181.8492	0.500335	0.5	2.120603	422	0.007033	0.712713
0.601809	0.402423	4256.03	0.687175	0.4	9.386935	1868	0.031133	0.80981
0.720037	0.284445	4944.859	0.572819	0.3	9.477387	1886	0.031433	0.70971
0.821793	0.182262	5639.844	0.522402	0.2	10.53266	2096	0.034933	0.642643
0.913883	0.089174	6579.03	0.487281	0.1	11.9397	2376	0.0396	0.601602
0.991707	0.008541	9056.206	0.466602	0.01	17.56281	3495	0.05825	0.525526
0.999181	8.44E-04	13391.93	0.469022	0.001	23.11055	4599	0.07665	0.535536
0.999917	8.39E-05	21918.47	0.479303	1.00E-04	37.1809	7399	0.123317	0.544545
0.999926	7.42E-05	22800.8	0.481295	9.00E-05	38.1206	7586	0.126433	0.536537
0.999934	6.62E-05	23006.81	0.480943	8.00E-05	38.9196	7745	0.129083	0.535536
0.999942	5.84E-05	23333.74	0.482115	7.00E-05	39.17588	7796	0.129933	0.531532
0.999951	4.97E-05	24551.22	0.482191	6.00E-05	41.08543	8176	0.136267	0.542543
0.999959	4.14E-05	26014.26	0.484072	5.00E-05	43.98995	8770	0.146167	0.541542
0.999968	3.26E-05	27481.4	0.483549	4.00E-05	45.69849	9094	0.151567	0.538539
0.999975	2.49E-05	29299.89	0.485491	3.00E-05	49.11055	9773	0.162883	0.558559
0.999984	1.64E-05	32683.28	0.488066	2.00E-05	54.68844	10899	0.18165	0.545546
0.999992	8.08E-06	40771.13	0.492367	1.00E-05	68.26131	13584	0.2264	0.565566
0.999993	7.23E-06	41611.21	0.492332	9.00E-06	69.76884	13900	0.231667	0.557558
0.999994	6.47E-06	44637.55	0.491688	8.00E-06	73.84422	14743	0.245717	0.544545
0.999994	5.73E-06	45775.19	0.492125	7.00E-06	76.01005	15126	0.2521	0.542543
0.999995	4.68E-06	48841.87	0.498227	6.00E-06	81.01508	16122	0.2687	0.573574
0.999996	4.01E-06	52807.21	0.499173	5.00E-06	87.58291	17429	0.290483	0.554555
0.999997	3.15E-06	57597.56	0.501099	4.00E-06	95.36683	18978	0.3163	0.617618
0.999998	2.39E-06	66097.34	0.506024	3.00E-06	109.4221	21775	0.362917	0.588589
0.999998	1.56E-06	89823.03	0.51373	2.00E-06	148.1357	29479	0.491317	0.596597
0.999999	7.50E-07	136550.8	0.526924	1.00E-06	224.8543	44746	0.745767	0.641642

训练集	0.5	c
	0.5	fa
测试集	0.6	c
	0.4	p	cfacp07
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	最大准确率p-max
0.49956	0.516006	181.0854	0.399304	0.5	2.201005	438	0.0073	0.511512
0.59172	0.411954	4191.769	0.710308	0.4	8.889447	1769	0.029483	0.847848
0.721064	0.28564	4952.065	0.630102	0.3	9.261307	1843	0.030717	0.73974
0.819327	0.184442	5590.94	0.586204	0.2	10.29648	2050	0.034167	0.695696
0.911947	0.089935	6354.437	0.566174	0.1	11.55779	2300	0.038333	0.623624
0.991821	0.008364	9071.04	0.549112	0.01	17.34171	3483	0.05805	0.593594
0.999194	8.24E-04	13482.6	0.552623	0.001	23.27638	4648	0.077467	0.603604
0.999919	8.14E-05	21949.47	0.563232	1.00E-04	36.89447	7342	0.122367	0.605606
0.999927	7.36E-05	22515.17	0.562678	9.00E-05	38.36181	7634	0.127233	0.603604
0.999934	6.71E-05	22916.88	0.566969	8.00E-05	39.21106	7803	0.13005	0.616617
0.999943	5.77E-05	23802.39	0.565661	7.00E-05	40.09548	7979	0.132983	0.61962
0.999951	4.97E-05	24917.23	0.565681	6.00E-05	42.22111	8418	0.1403	0.61962
0.999959	4.15E-05	26234.06	0.565098	5.00E-05	43.75377	8707	0.145117	0.613614
0.999968	3.26E-05	27612.55	0.568926	4.00E-05	46.22111	9230	0.153833	0.613614
0.999976	2.42E-05	29971.08	0.566521	3.00E-05	50.52261	10054	0.167567	0.604605
0.999984	1.61E-05	32603.34	0.571672	2.00E-05	54.69347	10885	0.181417	0.624625
0.999992	7.87E-06	40400.41	0.577075	1.00E-05	67.47236	13427	0.223783	0.635636
0.999993	7.24E-06	42048.5	0.579378	9.00E-06	70.32663	14027	0.233783	0.661662
0.999994	6.42E-06	44133.58	0.581516	8.00E-06	73.56281	14639	0.243983	0.651652
0.999994	5.62E-06	45537.41	0.58215	7.00E-06	76.01508	15143	0.252383	0.638639
0.999995	4.85E-06	49563.79	0.582422	6.00E-06	82.0201	16322	0.272033	0.625626
0.999996	4.02E-06	52837.96	0.584962	5.00E-06	88.23618	17560	0.292667	0.62963
0.999997	3.24E-06	57806.51	0.589821	4.00E-06	96.70854	19261	0.321017	0.653654
0.999998	2.35E-06	67227.74	0.594796	3.00E-06	111.8995	22268	0.371133	0.655656
0.999998	1.54E-06	89076.4	0.604066	2.00E-06	147.5427	29361	0.48935	0.67968
0.999999	7.61E-07	133486.9	0.614982	1.00E-06	221.206	44021	0.733683	0.682683

训练集	0.5	c
	0.5	fa
测试集	0.7	c
	0.3	p	cfacp08
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	最大准确率p-max
0.499559	0.516713	182.2915	0.301503	0.5	2.120603	453	0.00755	0.35035
0.599755	0.405568	4123.894	0.749589	0.4	10.20101	2030	0.033833	0.858859
0.718591	0.28422	5111.372	0.686762	0.3	10.67839	2125	0.035417	0.766767
0.82054	0.18506	5542.005	0.651853	0.2	10.45729	2081	0.034683	0.720721
0.91251	0.09052	6484.759	0.636179	0.1	11.74874	2338	0.038967	0.692693
0.991793	0.008363	9049.382	0.627089	0.01	16.30151	3244	0.054067	0.687688
0.999174	8.42E-04	13356.98	0.63392	0.001	22.80905	4539	0.07565	0.675676
0.999918	8.33E-05	21717.53	0.644962	1.00E-04	36.07538	7179	0.11965	0.6997
0.999927	7.39E-05	22438.03	0.645007	9.00E-05	38.01005	7564	0.126067	0.704705
0.999934	6.68E-05	23175.45	0.644252	8.00E-05	38.55276	7672	0.127867	0.684685
0.999943	5.78E-05	23993.19	0.644418	7.00E-05	40.11558	7999	0.133317	0.685686
0.999951	4.91E-05	25019.47	0.646129	6.00E-05	41.28141	8215	0.136917	0.685686
0.99996	4.07E-05	26405.5	0.649212	5.00E-05	43.8191	8720	0.145333	0.694695
0.999968	3.26E-05	27082.36	0.649831	4.00E-05	45.61809	9093	0.15155	0.694695
0.999975	2.49E-05	29145.53	0.651425	3.00E-05	48.46734	9677	0.161283	0.705706
0.999984	1.64E-05	33236.52	0.653895	2.00E-05	55.57789	11060	0.184333	0.692693
0.999992	7.99E-06	40474.13	0.664524	1.00E-05	66.99497	13332	0.2222	0.707708
0.999993	7.27E-06	42233.5	0.662728	9.00E-06	-236.643	-47092	-0.78487	0.701702
0.999994	6.51E-06	43326.19	0.666405	8.00E-06	70.67839	14081	0.234683	0.717718
0.999994	5.64E-06	44709.27	0.667014	7.00E-06	73.30653	14588	0.243133	0.70971
0.999995	4.84E-06	48636.44	0.66804	6.00E-06	79.92965	15906	0.2651	0.718719
0.999996	4.09E-06	51744.04	0.672839	5.00E-06	84.55276	16826	0.280433	0.716717
0.999997	3.22E-06	56998.86	0.67479	4.00E-06	94.36683	18794	0.313233	0.72973
0.999998	2.40E-06	68756.64	0.685741	3.00E-06	112.8995	22468	0.374467	0.74975
0.999998	1.54E-06	87909.9	0.694011	2.00E-06	144.4121	28738	0.478967	0.757758
0.999999	7.74E-07	131352.8	0.710645	1.00E-06	215.1005	42805	0.713417	0.780781

训练集	0.5	c
	0.5	fa
测试集	0.8	c
	0.2	p	cfacp09
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	最大准确率p-max
0.499581	0.517003	180.392	0.200487	0.5	2.211055	472	0.007867	0.293293
0.607987	0.397308	4235.276	0.776591	0.4	8.557789	1703	0.028383	0.860861
0.720443	0.283679	4976.578	0.740877	0.3	9.41206	1873	0.031217	0.817818
0.820826	0.18397	5547.141	0.723497	0.2	10.06533	2019	0.03365	0.785786
0.912978	0.089083	6474.176	0.707728	0.1	11.64824	2318	0.038633	0.767768
0.99165	0.008577	9066.327	0.705414	0.01	18.61307	3704	0.061733	0.768769
0.999164	8.51E-04	13411.05	0.715238	0.001	24.08543	4793	0.079883	0.771772
0.999918	8.33E-05	21821.92	0.727642	1.00E-04	37.0402	7371	0.12285	0.780781
0.999926	7.49E-05	22519.88	0.727396	9.00E-05	37.90452	7543	0.125717	0.777778
0.999933	6.73E-05	23153.5	0.729559	8.00E-05	39.06533	7774	0.129567	0.773774
0.999942	5.83E-05	23891.25	0.729383	7.00E-05	40.24623	8009	0.133483	0.773774
0.999951	4.96E-05	24543.87	0.732698	6.00E-05	40.98492	8156	0.135933	0.790791
0.999959	4.12E-05	26256.23	0.732245	5.00E-05	44.22111	8846	0.147433	0.780781
0.999967	3.36E-05	27262.95	0.734378	4.00E-05	45.45729	9046	0.150767	0.790791
0.999976	2.44E-05	30107.33	0.735982	3.00E-05	50.69849	10090	0.168167	0.791792
0.999984	1.63E-05	32933.07	0.73979	2.00E-05	55.19598	10984	0.183067	0.787788
0.999992	8.19E-06	40385.89	0.745982	1.00E-05	67.52261	13437	0.22395	0.798799
0.999993	7.36E-06	41868.91	0.74892	9.00E-06	69.31658	13826	0.230433	0.797798
0.999994	6.44E-06	44296.43	0.751007	8.00E-06	73.79397	14685	0.24475	0.786787
0.999994	5.74E-06	46905.53	0.752858	7.00E-06	77.88442	15499	0.258317	0.805806
0.999995	4.72E-06	49077.94	0.754247	6.00E-06	81.83417	16285	0.271417	0.810811
0.999996	3.99E-06	52174.19	0.757863	5.00E-06	86.81407	17276	0.287933	0.815816
0.999997	3.26E-06	57563.85	0.765534	4.00E-06	94.59799	18841	0.314017	0.808809
0.999998	2.37E-06	69164.3	0.772748	3.00E-06	114.4824	22782	0.3797	0.811812
0.999998	1.54E-06	88482.28	0.784362	2.00E-06	146.4573	29145	0.48575	0.826827
0.999999	7.44E-07	136206.8	0.80073	1.00E-06	224.1658	44609	0.743483	0.846847

训练集	0.5	c
	0.5	fa
测试集	0.9	c
	0.1	p	cfacp10
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	最大准确率p-max
0.499614	0.516383	182.0101	0.107218	0.5	2.276382	453	0.00755	0.996997
0.600948	0.404471	4116.653	0.80816	0.4	9.432161	1893	0.03155	0.916917
0.719438	0.284194	4989.271	0.798079	0.3	9.904523	1971	0.03285	0.861862
0.819831	0.183678	5747.281	0.783085	0.2	10.94472	2178	0.0363	0.851852
0.913547	0.088807	6409.07	0.7849	0.1	11.44221	2277	0.03795	0.837838
0.991835	0.008368	9216.95	0.783271	0.01	16.8191	3347	0.055783	0.85986
0.999201	8.20E-04	13467.39	0.800645	0.001	23.13568	4604	0.076733	0.856857
0.999917	8.38E-05	21907.3	0.808492	1.00E-04	36.97487	7358	0.122633	0.865866
0.999927	7.44E-05	22555.79	0.812164	9.00E-05	37.92965	7578	0.1263	0.878879
0.999933	6.77E-05	23243.63	0.810901	8.00E-05	39.16583	7794	0.1299	0.872873
0.999943	5.82E-05	24373.32	0.812687	7.00E-05	40.79397	8118	0.1353	0.866867
0.999952	4.87E-05	25032.78	0.810187	6.00E-05	42.86432	8530	0.142167	0.860861
0.999959	4.07E-05	25664.13	0.815791	5.00E-05	42.80402	8534	0.142233	0.866867
0.999967	3.32E-05	27545.23	0.8156	4.00E-05	45.8794	9145	0.152417	0.873874
0.999975	2.48E-05	29247.3	0.819845	3.00E-05	49.52261	9855	0.16425	0.872873
0.999984	1.65E-05	32611.02	0.82567	2.00E-05	54.52764	10867	0.181117	0.874875
0.999992	7.96E-06	39919.24	0.833623	1.00E-05	66.72864	13279	0.221317	0.884885
0.999993	7.19E-06	41749.18	0.836263	9.00E-06	69.25126	13796	0.229933	0.884885
0.999994	6.46E-06	43379.97	0.836605	8.00E-06	72.35176	14398	0.239967	0.891892
0.999994	5.61E-06	45526.25	0.838034	7.00E-06	75.49749	15040	0.250667	0.87988
0.999995	4.78E-06	47933.81	0.839563	6.00E-06	79.78894	15894	0.2649	0.882883
0.999996	3.99E-06	50998.32	0.846349	5.00E-06	84.49749	16815	0.28025	0.883884
0.999997	3.20E-06	58455.09	0.852863	4.00E-06	97.41709	19386	0.3231	0.901902
0.999998	2.42E-06	66801.87	0.863643	3.00E-06	110.8442	22058	0.367633	0.902903
0.999998	1.56E-06	91993.77	0.87739	2.00E-06	152.0704	30278	0.504633	0.916917
0.999999	7.48E-07	137608.4	0.89415	1.00E-06	226.9899	45171	0.75285	0.91992

训练集	0.5	c
	0.5	fa
测试集	1	c
	0	p	cfacp11
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	最大准确率p-max
0.499611	0.516635	179.8191	2.06E-04	0.5	2.040201	406	0.006767	0.041041
0.597441	0.407225	4231.196	0.83479	0.4	9.653266	1938	0.0323	0.924925
0.719142	0.286106	4992.678	0.85819	0.3	10.29648	2049	0.03415	0.923924
0.819366	0.184033	5551.503	0.854991	0.2	10.42714	2075	0.034583	0.935936
0.9123	0.088884	6358.397	0.854392	0.1	11.45226	2294	0.038233	0.921922
0.991748	0.008449	9023.151	0.866017	0.01	17.18593	3452	0.057533	0.944945
0.999191	8.26E-04	13362.45	0.878542	0.001	23.1407	4653	0.07755	0.951952
0.999918	8.31E-05	22031.74	0.890529	1.00E-04	37.57286	7477	0.124617	0.945946
0.999926	7.43E-05	22476.72	0.894739	9.00E-05	38.38191	7638	0.1273	0.947948
0.999934	6.74E-05	23198.58	0.894226	8.00E-05	40.37688	8066	0.134433	0.946947
0.999943	5.72E-05	23838.93	0.896208	7.00E-05	40.37186	8050	0.134167	0.964965
0.999951	4.89E-05	25013.68	0.896318	6.00E-05	42.47739	8484	0.1414	0.952953
0.999959	4.16E-05	26036.44	0.896218	5.00E-05	44.45226	8846	0.147433	0.957958
0.999967	3.29E-05	27582.12	0.900312	4.00E-05	47.0402	9377	0.156283	0.956957
0.999976	2.43E-05	29827.14	0.901469	3.00E-05	50.83417	10131	0.16885	0.962963
0.999984	1.64E-05	33400.63	0.906072	2.00E-05	56.99497	11342	0.189033	0.961962
0.999992	8.14E-06	40473.71	0.919759	1.00E-05	68.46231	13624	0.227067	0.964965
0.999993	7.19E-06	41492.56	0.920237	9.00E-06	70.42211	14014	0.233567	0.975976
0.999994	6.47E-06	42951.76	0.922209	8.00E-06	72.54271	14436	0.2406	0.961962
0.999994	5.71E-06	45793.2	0.922661	7.00E-06	77.42211	15407	0.256783	0.967968
0.999995	4.88E-06	47739.44	0.926434	6.00E-06	80.70352	16076	0.267933	0.974975
0.999996	3.99E-06	51814.72	0.933813	5.00E-06	88.08543	17529	0.29215	0.986987
0.999997	3.20E-06	57040.44	0.937978	4.00E-06	95.48241	19001	0.316683	0.98999
0.999998	2.34E-06	67561.62	0.95254	3.00E-06	112.6884	22440	0.374	0.994995
0.999998	1.58E-06	87890.54	0.965639	2.00E-06	146.593	29172	0.4862	1
0.999999	7.59E-07	134455	0.984256	1.00E-06	223.201	44417	0.740283	1

黑榆

关注

0
点赞
踩
0

收藏

觉得还不错? 一键收藏
打赏
0
评论
ANN神经网络分类2*2矩阵：吸引子和反鞍点cfa-cp

用吸引子和反鞍点做训练集用排斥子和吸引子做测试集，通过改变测试集吸引子和排斥子的比例观察分类效果，并总结规律。 r1 r2 <1 <1 吸引子 c >1 >1 排斥子 ...
复制链接

扫一扫