3层、5层、3层一个卷积核BP神经网络性能比较
3層網絡的結構是
d2（mnist 0,1）-49-30-2-（2*k） ,k∈(0,1)
分類mnist的0和1，三層的節點數分別是49，30，2激活函數用sigmoid。讓0向1，0收斂讓1向0，1收斂。
?
5層網絡的結構是
d2（mnist 0,1）-81-30-49-30-2-（2*k） ,k∈(0,1)
分類mnist的0和1，五層的節點數分別是81，30，49，30，2激活函數用sigmoid。讓0向1，0收斂讓1向0，1收斂。
?
3層一個卷積核的網絡的結構是
d2（mnist 0,1）81-con（3*3）49-30-2-（2*k） ,k∈(0,1)
分類mnist的0和1，輸入9*9的圖片經過一個3*3的卷積核尺寸變成7*7，隱藏層30個節點輸出層2個節點。激活函數用sigmoid。讓0向1，0收斂讓1向0，1收斂。
?
具體進樣順序
d2（mnist 0,1）81-con（3*3）49-30-2-（2*k） ,k∈(0,1)
對應的網絡結構
這個網絡的收斂標準是
if (Math.abs(f2[0]-y[0])< δ? &&? Math.abs(f2[1]-y[1])< δ?? )
| 具體進樣順序 | ? | ? | ? |
| δ=0.5 | 迭代次數 | ? | ? |
| minst 0-1 | 1 | 判斷是否達到收斂 | |
| minst 1-1 | 2 | 判斷是否達到收斂 | |
| 梯度下降 | ? | ? | ? |
| minst 0-2 | 3 | 判斷是否達到收斂 | |
| minst 1-2 | 4 | 判斷是否達到收斂 | |
| …… | ? | ? | ? |
| minst 0-4999 | 9997 | 判斷是否達到收斂 | |
| minst 1-4999 | 9998 | 判斷是否達到收斂 | |
| 梯度下降 | ? | ? | ? |
| …… | ? | ? | ? |
| 如果4999圖片內沒有達到收斂標準再次從頭循環 | |||
| minst 0-1 | 9999 | 判斷是否達到收斂 | |
| minst 1-1 | 10000 | 判斷是否達到收斂 | |
| …… | ? | ? | ? |
| 每當網路達到收斂標準記錄迭代次數和對應的準確率測試結果 | |||
| 將這一過程重復199次 | ? | ? | |
| δ=0.01 | ? | ? | ? |
| …… | ? | ? | ? |
| δ=1e-6 | ? | ? | ? |
?
對應每個收斂標準都要收斂199次取平均值，記錄199次的最大值，嘗試了從0.5到1e-6共34個收斂標準，所以對應每個網絡需收斂199*34次。
通過比較平均準確率，最大準確率，迭代次數，收斂時間比較三個網絡的性能
| ? | 3層 | 5層 | 3層1個卷積核 | ? | ? | 3層 | 5層 | 3層1個卷積核 |
| δ | 平均準確率p-ave | 平均準確率p-ave | 平均準確率p-ave | δ | 最大值p-max | 最大值p-max | 最大值p-max | |
| 0.5 | 0.556121031 | 0.502438908 | 0.517770888 | ? | 0.5 | 0.941371158 | 0.810874704 | 0.867612293 |
| 0.4 | 0.985338038 | 0.906190527 | 0.75949725 | ? | 0.4 | 0.992434988 | 0.997163121 | 0.996690307 |
| 0.3 | 0.988685746 | 0.99093339 | 0.914275871 | ? | 0.3 | 0.991489362 | 0.99858156 | 0.998108747 |
| 0.2 | 0.988645355 | 0.996196111 | 0.944265061 | ? | 0.2 | 0.990543735 | 0.999054374 | 0.998108747 |
| 0.1 | 0.987623698 | 0.996065434 | 0.953538377 | ? | 0.1 | 0.990543735 | 0.999054374 | 0.997635934 |
| 0.01 | 0.989999644 | 0.995502334 | 0.928536298 | ? | 0.01 | 0.991489362 | 0.999054374 | 0.998108747 |
| 0.001 | 0.968157573 | 0.989992516 | 0.907932095 | ? | 0.001 | 0.992907801 | 1 | 0.99858156 |
| 9.00E-04 | 0.968585243 | 0.991353933 | 0.890459389 | ? | 9.00E-04 | 0.992907801 | 1 | 0.99858156 |
| 8.00E-04 | 0.972377253 | 0.989754921 | 0.901728501 | ? | 8.00E-04 | 0.992907801 | 0.999527187 | 0.998108747 |
| 7.00E-04 | 0.978036756 | 0.990120817 | 0.897024128 | ? | 7.00E-04 | 0.992907801 | 1 | 0.997635934 |
| 6.00E-04 | 0.985468715 | 0.99272723 | 0.898440191 | ? | 6.00E-04 | 0.993380615 | 0.999527187 | 0.999527187 |
| 5.00E-04 | 0.991769723 | 0.991821994 | 0.881307245 | ? | 5.00E-04 | 0.992907801 | 0.999527187 | 0.998108747 |
| 4.00E-04 | 0.992444492 | 0.994896468 | 0.872397448 | ? | 4.00E-04 | 0.993380615 | 0.999527187 | 0.998108747 |
| 3.00E-04 | 0.992969576 | 0.995005762 | 0.884989962 | ? | 3.00E-04 | 0.994799054 | 0.999527187 | 0.999054374 |
| 2.00E-04 | 0.994399895 | 0.994136165 | 0.887848225 | ? | 2.00E-04 | 0.994799054 | 0.998108747 | 0.998108747 |
| 1.00E-04 | 0.993960346 | 0.991389572 | 0.847616332 | ? | 1.00E-04 | 0.994799054 | 1 | 0.999054374 |
| 9.00E-05 | 0.993421006 | 0.993760766 | 0.870898226 | ? | 9.00E-05 | 0.994799054 | 0.999527187 | 0.99858156 |
| 8.00E-05 | 0.993409126 | 0.994818062 | 0.846815638 | ? | 8.00E-05 | 0.994799054 | 0.999527187 | 0.999527187 |
| 7.00E-05 | 0.993663352 | 0.996609525 | 0.858384119 | ? | 7.00E-05 | 0.994799054 | 0.999527187 | 0.99858156 |
| 6.00E-05 | 0.99426209 | 0.996367179 | 0.822433681 | ? | 6.00E-05 | 0.994799054 | 0.999527187 | 0.999054374 |
| 5.00E-05 | 0.994269218 | 0.992064341 | 0.812162467 | ? | 5.00E-05 | 0.994799054 | 0.999527187 | 0.99858156 |
| 4.00E-05 | 0.99394609 | 0.988716633 | 0.833256115 | ? | 4.00E-05 | 0.994799054 | 0.999054374 | 0.999527187 |
| 3.00E-05 | 0.993435262 | 0.987889804 | 0.860681659 | ? | 3.00E-05 | 0.993853428 | 0.997163121 | 0.999054374 |
| 2.00E-05 | 0.993437638 | 0.990051915 | 0.861140216 | ? | 2.00E-05 | 0.994326241 | 0.996690307 | 0.999054374 |
| 1.00E-05 | 0.994366632 | 0.989160935 | 0.868161137 | ? | 1.00E-05 | 0.995271868 | 0.994326241 | 0.998108747 |
| 9.00E-06 | 0.994266842 | 0.989065897 | 0.856100835 | ? | 9.00E-06 | 0.994799054 | 0.992907801 | 0.999054374 |
| 8.00E-06 | 0.993960346 | 0.989201326 | 0.856908657 | ? | 8.00E-06 | 0.995271868 | 0.992434988 | 0.999054374 |
| 7.00E-06 | 0.993810661 | 0.989217957 | 0.851536643 | ? | 7.00E-06 | 0.994799054 | 0.992907801 | 0.999054374 |
| 6.00E-06 | 0.994031624 | 0.989120544 | 0.835860152 | ? | 6.00E-06 | 0.994799054 | 0.992434988 | 0.999527187 |
| 5.00E-06 | 0.994314361 | 0.988958979 | 0.839217363 | ? | 5.00E-06 | 0.994799054 | 0.992907801 | 0.999527187 |
| 4.00E-06 | 0.993972225 | 0.988844934 | 0.82416337 | ? | 4.00E-06 | 0.994799054 | 0.994799054 | 0.999054374 |
| 3.00E-06 | 0.993385367 | 0.989270228 | 0.84103021 | ? | 3.00E-06 | 0.993853428 | 0.994799054 | 0.999054374 |
| 2.00E-06 | 0.993836796 | 0.99128503 | 0.857692719 | ? | 2.00E-06 | 0.993853428 | 0.998108747 | 0.999054374 |
| 1.00E-06 | 0.993896195 | 0.991068819 | 0.887536976 | ? | 1.00E-06 | 0.994326241 | 0.99858156 | 0.999054374 |
?
199次平均準確率p-ave
| δ=1e-6 | 3層 | > | 5層 | > | 3層1個卷積核 |
| ? | 1.11983638 | ? | 1.116650738 | ? | 1 |
?
也就是49-30-2的網絡結構上加一個3*3的核導致平均性能下降，
在49-30-2的結構上加兩層81*30也會導致平均性能下降，但要好于加卷積核。
199次的最大準確率p-max
| δ=1e-6 | 3層1個卷積核 | > | 5層 | > | 3層 |
| ? | 1.004755112 | ? | 1.0042796 | ? | 1 |
?在49-30-2的基礎上加一個卷積核非常明顯的提高了網絡的最大分辨率。在49-30-2的基礎上加兩層81-30也可以提高分辨率但是沒有加卷積核效果好。
綜合前兩個表
加卷積核和加層數確實都會使最大性能上升，而且針對測試的3個網絡加卷積核的效果更明顯。
| ? | 3層 | 5層 | 3層1個卷積核 | ? | ? | 3層 | 5層 | 3層1個卷積核 |
| δ | 迭代次數n | 迭代次數n | 迭代次數n | ? | δ | 耗時 min/199 | 耗時 min/199 | 耗時 min/199 |
| 0.5 | 9.457286432 | 17.31155779 | 16.48241206 | ? | 0.5 | 4.0988 | 4.0001 | 2.255266667 |
| 0.4 | 217.8743719 | 3767.874372 | 1361.045226 | ? | 0.4 | 4.409 | 17.10336667 | 3.560683333 |
| 0.3 | 282.9095477 | 3700.497487 | 1720.190955 | ? | 0.3 | 4.53095 | 16.88956667 | 3.8671 |
| 0.2 | 359.3015075 | 3768.60804 | 1943.336683 | ? | 0.2 | 4.715983333 | 20.48565 | 4.079916667 |
| 0.1 | 440.5829146 | 3784.150754 | 2060.557789 | ? | 0.1 | 1.64245 | 22.22303333 | 4.198666667 |
| 0.01 | 898.4974874 | 3937.753769 | 2972.638191 | ? | 0.01 | 5.714416667 | 22.55365 | 5.0917 |
| 0.001 | 1712.040201 | 4656.899497 | 4091.859296 | ? | 0.001 | 7.16 | 25.62353333 | 6.158 |
| 9.00E-04 | 1735.075377 | 4732.698492 | 4177.100503 | ? | 9.00E-04 | 7.092683333 | 26.34898333 | 6.25685 |
| 8.00E-04 | 1817.145729 | 4773.040201 | 4204.231156 | ? | 8.00E-04 | 7.23665 | 26.55536667 | 6.278333333 |
| 7.00E-04 | 1967.728643 | 4867.150754 | 4327.341709 | ? | 7.00E-04 | 7.516983333 | 28.6609 | 6.3886 |
| 6.00E-04 | 2466.407035 | 5051.743719 | 4319.231156 | ? | 6.00E-04 | 8.460533333 | 25.1193 | 6.332716667 |
| 5.00E-04 | 2922.668342 | 5123.552764 | 4640.180905 | ? | 5.00E-04 | 9.2511 | 28.59831667 | 6.635683333 |
| 4.00E-04 | 2991.839196 | 5354.537688 | 4781.437186 | ? | 4.00E-04 | 9.083083333 | 27.10175 | 6.873116667 |
| 3.00E-04 | 4905.969849 | 5771.562814 | 4958.276382 | ? | 3.00E-04 | 12.79875 | 31.4649 | 7.17385 |
| 2.00E-04 | 5184 | 6608.065327 | 5410.175879 | ? | 2.00E-04 | 13.1458 | 33.98488333 | 7.57195 |
| 1.00E-04 | 5632.281407 | 9968.100503 | 5985.060302 | ? | 1.00E-04 | 14.19561667 | 49.20433333 | 9.430166667 |
| 9.00E-05 | 5723.949749 | 10986.21608 | 5960.879397 | ? | 9.00E-05 | 14.03463333 | 53.35168333 | 7.974133333 |
| 8.00E-05 | 5754.090452 | 11628.0201 | 6234.261307 | ? | 8.00E-05 | 14.26598333 | 56.16275 | 9.804233333 |
| 7.00E-05 | 6029.919598 | 12679.52764 | 6227.025126 | ? | 7.00E-05 | 14.79435 | 61.09628333 | 9.777633333 |
| 6.00E-05 | 6686.894472 | 13109.78894 | 6410.015075 | ? | 6.00E-05 | 15.92916667 | 63.26216667 | 10.05916667 |
| 5.00E-05 | 6733.869347 | 13711.12563 | 6843.527638 | ? | 5.00E-05 | 15.9815 | 65.85913333 | 10.4674 |
| 4.00E-05 | 7164.261307 | 15135.42211 | 7226.582915 | ? | 4.00E-05 | 17.0012 | 71.99621667 | 10.87256667 |
| 3.00E-05 | 7876 | 20739.68844 | 7567.170854 | ? | 3.00E-05 | 18.42911667 | 97.38201667 | 11.22465 |
| 2.00E-05 | 7898.773869 | 33675.49246 | 8543.718593 | ? | 2.00E-05 | 18.3598 | 134.5200333 | 12.26171667 |
| 1.00E-05 | 10519.22613 | 56278.17085 | 10002.80905 | ? | 1.00E-05 | 20.59985 | 249.1469333 | 13.94281667 |
| 9.00E-06 | 10562.28141 | 59442.55276 | 10314.70854 | ? | 9.00E-06 | 23.99378333 | 262.6644167 | 14.27883333 |
| 8.00E-06 | 10718.19095 | 62130.47739 | 10521.40704 | ? | 8.00E-06 | 24.3037 | 276.2772 | 14.46355 |
| 7.00E-06 | 10886.38191 | 66274.23618 | 10795.59296 | ? | 7.00E-06 | 23.87303333 | 232.7295667 | 15.33916667 |
| 6.00E-06 | 11353.8392 | 70016.88442 | 11356.43216 | ? | 6.00E-06 | 25.2186 | 243.7804833 | 16.13403333 |
| 5.00E-06 | 13691.06533 | 75014.68844 | 11524.1407 | ? | 5.00E-06 | 29.18276667 | 262.0030333 | 16.73851667 |
| 4.00E-06 | 16101.14573 | 81911.47739 | 12755.88945 | ? | 4.00E-06 | 30.34141667 | 283.9857167 | 18.1965 |
| 3.00E-06 | 16729.59799 | 92364.45729 | 13319.76884 | ? | 3.00E-06 | 34.85281667 | 319.9373 | 18.90775 |
| 2.00E-06 | 17837.20603 | 113501.8392 | 15362.49749 | ? | 2.00E-06 | 32.95221667 | 392.8333667 | 21.25796667 |
| 1.00E-06 | 20745.63819 | 154350.5126 | 17225.44221 | ? | 1.00E-06 | 41.03785 | 559.2056333 | 23.45475 |
?
迭代次數
| δ=1e-6 | 5層 | > | 3層 | > | 3層一個卷積核 |
| ? | 8.96061249 | ? | 1.204360268 | ? | 1 |
5層網絡的迭代次數要遠大于其他的兩個網絡。
?
耗時
| δ=1e-6 | 5層 | > | 3層 | > | 3層一個卷積核 |
| ? | 23.84189272 | ? | 1.749660517 | ? | 1 |
5層網絡消耗了23倍的時間取得的最大準確率仍小于3層一個卷積核的網絡。
?
?
199次平均準確率p-ave
3層>5層>3層1個卷積核
199次的最大準確率p-max
3層1個卷積核>5層>3層
迭代次數
5層>3層>3層一個卷積核
耗時
5層>3層>3層一個卷積核
所以針對測試的3個網絡在49-30-2的基礎上加兩層81-30會使網絡的平均性能基本不變的情況下最大性能顯著提升，但代價是收斂速度嚴重下降。
在49-30-2的基礎上增加卷積核會大幅提升網絡的最大性能，但卷積核太少了會影響網絡的平均性能，使性能變得不穩定。
?
?
| 實驗數據 |
| 學習率 0.1 |
| 權重初始化方式 |
| 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); |
| 第一層第二層和卷積核的權重的初始化的x分別為1000,1000,200 |
3層網絡的2層權重x=1000
5層網絡的4層權重x=1000
3層1個卷積核網絡的2層權重x=1000，卷積核的x=200
?
| 49-30-2 | ? | ? | ? | ? | ? | ? | ? | ? |
| f2[0] | f2[1] | 迭代次數n | 平均準確率p-ave | δ | 耗時ms/次 | 耗時ms/199次 | 耗時 min/199 | 最大值p-max |
| 0.499386185 | 0.499447667 | 9.457286432 | 0.556121031 | 0.5 | 1235.738693 | 245928 | 4.0988 | 0.941371158 |
| 0.607789346 | 0.391697375 | 217.8743719 | 0.985338038 | 0.4 | 1329.266332 | 264540 | 4.409 | 0.992434988 |
| 0.717279441 | 0.282649246 | 282.9095477 | 0.988685746 | 0.3 | 1366.035176 | 271857 | 4.53095 | 0.991489362 |
| 0.80371583 | 0.196464405 | 359.3015075 | 0.988645355 | 0.2 | 1421.899497 | 282959 | 4.715983333 | 0.990543735 |
| 0.845146872 | 0.154991397 | 440.5829146 | 0.987623698 | 0.1 | 495.1306533 | 98547 | 1.64245 | 0.990543735 |
| 0.081706915 | 0.918288828 | 898.4974874 | 0.989999644 | 0.01 | 1722.939698 | 342865 | 5.714416667 | 0.991489362 |
| 0.974083268 | 0.025916326 | 1712.040201 | 0.968157573 | 0.001 | 2158.79397 | 429600 | 7.16 | 0.992907801 |
| 0.923964157 | 0.07603503 | 1735.075377 | 0.968585243 | 9.00E-04 | 2138.497487 | 425561 | 7.092683333 | 0.992907801 |
| 0.753411266 | 0.246588817 | 1817.145729 | 0.972377253 | 8.00E-04 | 2181.904523 | 434199 | 7.23665 | 0.992907801 |
| 0.532621804 | 0.467378406 | 1967.728643 | 0.978036756 | 7.00E-04 | 2266.427136 | 451019 | 7.516983333 | 0.992907801 |
| 0.266563765 | 0.733435398 | 2466.407035 | 0.985468715 | 6.00E-04 | 2550.909548 | 507632 | 8.460533333 | 0.993380615 |
| 0.025442755 | 0.974557558 | 2922.668342 | 0.991769723 | 5.00E-04 | 2789.276382 | 555066 | 9.2511 | 0.992907801 |
| 3.34E-04 | 0.999666356 | 2991.839196 | 0.992444492 | 4.00E-04 | 2738.61809 | 544985 | 9.083083333 | 0.993380615 |
| 2.61E-04 | 0.999738897 | 4905.969849 | 0.992969576 | 3.00E-04 | 3858.919598 | 767925 | 12.79875 | 0.994799054 |
| 1.13E-04 | 0.999886811 | 5184 | 0.994399895 | 2.00E-04 | 3963.472362 | 788748 | 13.1458 | 0.994799054 |
| 8.53E-05 | 0.99991484 | 5632.281407 | 0.993960346 | 1.00E-04 | 4280.080402 | 851737 | 14.19561667 | 0.994799054 |
| 6.87E-05 | 0.999931338 | 5723.949749 | 0.993421006 | 9.00E-05 | 4231.547739 | 842078 | 14.03463333 | 0.994799054 |
| 6.72E-05 | 0.999932802 | 5754.090452 | 0.993409126 | 8.00E-05 | 4301.301508 | 855959 | 14.26598333 | 0.994799054 |
| 5.81E-05 | 0.99994191 | 6029.919598 | 0.993663352 | 7.00E-05 | 4460.60804 | 887661 | 14.79435 | 0.994799054 |
| 4.07E-05 | 0.999959309 | 6686.894472 | 0.99426209 | 6.00E-05 | 4802.758794 | 955750 | 15.92916667 | 0.994799054 |
| 3.96E-05 | 0.999960391 | 6733.869347 | 0.994269218 | 5.00E-05 | 4818.361809 | 958890 | 15.9815 | 0.994799054 |
| 2.97E-05 | 0.999970307 | 7164.261307 | 0.99394609 | 4.00E-05 | 5125.984925 | 1020072 | 17.0012 | 0.994799054 |
| 1.52E-05 | 0.999984755 | 7876 | 0.993435262 | 3.00E-05 | 5556.512563 | 1105747 | 18.42911667 | 0.993853428 |
| 1.50E-05 | 0.999984982 | 7898.773869 | 0.993437638 | 2.00E-05 | 5535.613065 | 1101588 | 18.3598 | 0.994326241 |
| 8.24E-06 | 0.999991749 | 10519.22613 | 0.994366632 | 1.00E-05 | 6211.005025 | 1235991 | 20.59985 | 0.995271868 |
| 7.83E-06 | 0.99999217 | 10562.28141 | 0.994266842 | 9.00E-06 | 7234.301508 | 1439627 | 23.99378333 | 0.994799054 |
| 6.59E-06 | 0.999993413 | 10718.19095 | 0.993960346 | 8.00E-06 | 7327.748744 | 1458222 | 24.3037 | 0.995271868 |
| 5.83E-06 | 0.999994177 | 10886.38191 | 0.993810661 | 7.00E-06 | 7197.854271 | 1432382 | 23.87303333 | 0.994799054 |
| 5.41E-06 | 0.999994592 | 11353.8392 | 0.994031624 | 6.00E-06 | 7603.512563 | 1513116 | 25.2186 | 0.994799054 |
| 4.47E-06 | 0.999995529 | 13691.06533 | 0.994314361 | 5.00E-06 | 8798.658291 | 1750966 | 29.18276667 | 0.994799054 |
| 3.26E-06 | 0.999996747 | 16101.14573 | 0.993972225 | 4.00E-06 | 9148.165829 | 1820485 | 30.34141667 | 0.994799054 |
| 2.31E-06 | 0.999997686 | 16729.59799 | 0.993385367 | 3.00E-06 | 10508.29648 | 2091169 | 34.85281667 | 0.993853428 |
| 1.14E-06 | 0.999998862 | 17837.20603 | 0.993836796 | 2.00E-06 | 9935.180905 | 1977133 | 32.95221667 | 0.993853428 |
| 8.60E-07 | 0.999999139 | 20745.63819 | 0.993896195 | 1.00E-06 | 12373.14573 | 2462271 | 41.03785 | 0.994326241 |
| 7.97E-07 | 0.999999204 | 21273.60804 | 0.993903323 | 9.00E-07 | 12732.64322 | 2533802 | 42.23003333 | 0.994799054 |
| 6.87E-07 | 0.999999313 | 22264.36181 | 0.994050631 | 8.00E-07 | 13595.05025 | 2705417 | 45.09028333 | 0.994799054 |
| 6.10E-07 | 0.99999939 | 23553.17588 | 0.993943714 | 7.00E-07 | 13737.59799 | 2733784 | 45.56306667 | 0.994799054 |
| 4.87E-07 | 0.999999513 | 26313.60804 | 0.993470901 | 6.00E-07 | 15162.24623 | 3017303 | 50.28838333 | 0.994799054 |
| 4.06E-07 | 0.999999593 | 27056.1407 | 0.993577818 | 5.00E-07 | 14951.58291 | 2975365 | 49.58941667 | 0.994799054 |
| 2.56E-07 | 0.999999744 | 27868 | 0.994055383 | 4.00E-07 | 15827.96482 | 3149765 | 52.49608333 | 0.994799054 |
| 2.54E-07 | 0.999999745 | 27998.54271 | 0.994026872 | 3.00E-07 | 15686.69347 | 3121656 | 52.0276 | 0.994799054 |
| 1.72E-07 | 0.999999827 | 34769.20603 | 0.99349466 | 2.00E-07 | 19260.24623 | 3832821 | 63.88035 | 0.995271868 |
| 8.67E-08 | 0.999999913 | 38491.84925 | 0.994183684 | 1.00E-07 | 21997.96985 | 4377613 | 72.96021667 | 0.994799054 |
?
?
?
| 81*30*49*30*2 | ? | ? | ? | ? | ? | ? | ? | |
| f2[0] | f2[1] | 迭代次數n | 平均準確率p-ave | δ | 耗時ms/次 | 耗時ms/199次 | 耗時 min/199 | 最大值p-max |
| 0.497397008 | 0.501220403 | 17.31155779 | 0.502438908 | 0.5 | 1205.964824 | 240006 | 4.0001 | 0.810874704 |
| 0.58726634 | 0.412611563 | 3767.874372 | 0.906190527 | 0.4 | 5156.79397 | 1026202 | 17.10336667 | 0.997163121 |
| 0.636640566 | 0.36278485 | 3700.497487 | 0.99093339 | 0.3 | 5092.331658 | 1013374 | 16.88956667 | 0.99858156 |
| 0.654326877 | 0.345588026 | 3768.60804 | 0.996196111 | 0.2 | 6176.577889 | 1229139 | 20.48565 | 0.999054374 |
| 0.700519621 | 0.299452945 | 3784.150754 | 0.996065434 | 0.1 | 6700.40201 | 1333382 | 22.22303333 | 0.999054374 |
| 0.532134869 | 0.467869686 | 3937.753769 | 0.995502334 | 0.01 | 6800.080402 | 1353219 | 22.55365 | 0.999054374 |
| 0.422209109 | 0.577793465 | 4656.899497 | 0.989992516 | 0.001 | 7725.658291 | 1537412 | 25.62353333 | 1 |
| 0.341900165 | 0.65810115 | 4732.698492 | 0.991353933 | 9.00E-04 | 7944.38191 | 1580939 | 26.34898333 | 1 |
| 0.341879526 | 0.658120602 | 4773.040201 | 0.989754921 | 8.00E-04 | 8006.633166 | 1593322 | 26.55536667 | 0.999527187 |
| 0.296683534 | 0.703317013 | 4867.150754 | 0.990120817 | 7.00E-04 | 8641.472362 | 1719654 | 28.6609 | 1 |
| 0.186211679 | 0.813787865 | 5051.743719 | 0.99272723 | 6.00E-04 | 7573.653266 | 1507158 | 25.1193 | 0.999527187 |
| 0.201214859 | 0.798784971 | 5123.552764 | 0.991821994 | 5.00E-04 | 8622.60804 | 1715899 | 28.59831667 | 0.999527187 |
| 0.105760421 | 0.894239063 | 5354.537688 | 0.994896468 | 4.00E-04 | 8171.38191 | 1626105 | 27.10175 | 0.999527187 |
| 0.07558383 | 0.924415606 | 5771.562814 | 0.995005762 | 3.00E-04 | 9486.904523 | 1887894 | 31.4649 | 0.999527187 |
| 0.251341598 | 0.748658773 | 6608.065327 | 0.994136165 | 2.00E-04 | 10246.69849 | 2039093 | 33.98488333 | 0.998108747 |
| 0.964739483 | 0.035260641 | 9968.100503 | 0.991389572 | 1.00E-04 | 14835.39196 | 2952260 | 49.20433333 | 1 |
| 0.979821643 | 0.020178336 | 10986.21608 | 0.993760766 | 9.00E-05 | 16085.84925 | 3201101 | 53.35168333 | 0.999527187 |
| 0.994905063 | 0.005094859 | 11628.0201 | 0.994818062 | 8.00E-05 | 16933.49246 | 3369765 | 56.16275 | 0.999527187 |
| 0.999940459 | 5.95E-05 | 12679.52764 | 0.996609525 | 7.00E-05 | 18420.98995 | 3665777 | 61.09628333 | 0.999527187 |
| 0.999949295 | 5.05E-05 | 13109.78894 | 0.996367179 | 6.00E-05 | 19074.0201 | 3795730 | 63.26216667 | 0.999527187 |
| 0.999958601 | 4.14E-05 | 13711.12563 | 0.992064341 | 5.00E-05 | 19857.02513 | 3951548 | 65.85913333 | 0.999527187 |
| 0.999966141 | 3.39E-05 | 15135.42211 | 0.988716633 | 4.00E-05 | 21707.40201 | 4319773 | 71.99621667 | 0.999054374 |
| 0.999973419 | 2.65E-05 | 20739.68844 | 0.987889804 | 3.00E-05 | 29361.32663 | 5842921 | 97.38201667 | 0.997163121 |
| 0.999981476 | 1.85E-05 | 33675.49246 | 0.990051915 | 2.00E-05 | 40558.68342 | 8071202 | 134.5200333 | 0.996690307 |
| 0.999990802 | 9.20E-06 | 56278.17085 | 0.989160935 | 1.00E-05 | 75119.59799 | 14948816 | 249.1469333 | 0.994326241 |
| 0.999991844 | 8.16E-06 | 59442.55276 | 0.989065897 | 9.00E-06 | 79195.29648 | 15759865 | 262.6644167 | 0.992907801 |
| 0.999992753 | 7.24E-06 | 62130.47739 | 0.989201326 | 8.00E-06 | 83299.57789 | 16576632 | 276.2772 | 0.992434988 |
| 0.999993663 | 6.33E-06 | 66274.23618 | 0.989217957 | 7.00E-06 | 70169.71357 | 13963774 | 232.7295667 | 0.992907801 |
| 0.999994542 | 5.46E-06 | 70016.88442 | 0.989120544 | 6.00E-06 | 73501.49246 | 14626829 | 243.7804833 | 0.992434988 |
| 0.999995425 | 4.58E-06 | 75014.68844 | 0.988958979 | 5.00E-06 | 78995.73367 | 15720182 | 262.0030333 | 0.992907801 |
| 0.999996283 | 3.72E-06 | 81911.47739 | 0.988844934 | 4.00E-06 | 85623.82915 | 17039143 | 283.9857167 | 0.994799054 |
| 0.999997181 | 2.82E-06 | 92364.45729 | 0.989270228 | 3.00E-06 | 96463.50754 | 19196238 | 319.9373 | 0.994799054 |
| 0.999998086 | 1.91E-06 | 113501.8392 | 0.99128503 | 2.00E-06 | 118442.201 | 23570002 | 392.8333667 | 0.998108747 |
| 0.999999045 | 9.56E-07 | 154350.5126 | 0.991068819 | 1.00E-06 | 168604.7085 | 33552338 | 559.2056333 | 0.99858156 |
?
?
| con（3*3）-49-30-2 | ? | ? | ? | ? | ? | ? | ? | |
| f2[0] | f2[1] | 迭代次數n | 平均準確率p-ave | δ | 耗時ms/次 | 耗時ms/199次 | 耗時 min/199 | 最大值p-max |
| 0.500276107 | 0.498344763 | 16.48241206 | 0.517770888 | 0.5 | 679.9798995 | 135316 | 2.255266667 | 0.867612293 |
| 0.5818015 | 0.418241698 | 1361.045226 | 0.75949725 | 0.4 | 1073.572864 | 213641 | 3.560683333 | 0.996690307 |
| 0.644166219 | 0.35596743 | 1720.190955 | 0.914275871 | 0.3 | 1165.959799 | 232026 | 3.8671 | 0.998108747 |
| 0.725259837 | 0.274933998 | 1943.336683 | 0.944265061 | 0.2 | 1230.040201 | 244795 | 4.079916667 | 0.998108747 |
| 0.814017223 | 0.186277612 | 2060.557789 | 0.953538377 | 0.1 | 1265.929648 | 251920 | 4.198666667 | 0.997635934 |
| 0.873548828 | 0.126458355 | 2972.638191 | 0.928536298 | 0.01 | 1535.18593 | 305502 | 5.0917 | 0.998108747 |
| 0.873795911 | 0.12620395 | 4091.859296 | 0.907932095 | 0.001 | 1856.603015 | 369480 | 6.158 | 0.99858156 |
| 0.868847942 | 0.131153585 | 4177.100503 | 0.890459389 | 9.00E-04 | 1886.482412 | 375411 | 6.25685 | 0.99858156 |
| 0.883973522 | 0.116025939 | 4204.231156 | 0.901728501 | 8.00E-04 | 1892.964824 | 376700 | 6.278333333 | 0.998108747 |
| 0.858927553 | 0.141072196 | 4327.341709 | 0.897024128 | 7.00E-04 | 1926.211055 | 383316 | 6.3886 | 0.997635934 |
| 0.879064535 | 0.120935924 | 4319.231156 | 0.898440191 | 6.00E-04 | 1909.361809 | 379963 | 6.332716667 | 0.999527187 |
| 0.843952852 | 0.156046265 | 4640.180905 | 0.881307245 | 5.00E-04 | 2000.708543 | 398141 | 6.635683333 | 0.998108747 |
| 0.823934671 | 0.17606553 | 4781.437186 | 0.872397448 | 4.00E-04 | 2072.291457 | 412387 | 6.873116667 | 0.998108747 |
| 0.879229957 | 0.120769961 | 4958.276382 | 0.884989962 | 3.00E-04 | 2162.969849 | 430431 | 7.17385 | 0.999054374 |
| 0.819003562 | 0.180996744 | 5410.175879 | 0.887848225 | 2.00E-04 | 2282.919598 | 454317 | 7.57195 | 0.998108747 |
| 0.859247532 | 0.1407525 | 5985.060302 | 0.847616332 | 1.00E-04 | 2843.241206 | 565810 | 9.430166667 | 0.999054374 |
| 0.829105325 | 0.170894589 | 5960.879397 | 0.870898226 | 9.00E-05 | 2404.231156 | 478448 | 7.974133333 | 0.99858156 |
| 0.824083387 | 0.17591666 | 6234.261307 | 0.846815638 | 8.00E-05 | 2956.020101 | 588254 | 9.804233333 | 0.999527187 |
| 0.864285382 | 0.135714544 | 6227.025126 | 0.858384119 | 7.00E-05 | 2947.98995 | 586658 | 9.777633333 | 0.99858156 |
| 0.909514101 | 0.090486022 | 6410.015075 | 0.822433681 | 6.00E-05 | 3032.884422 | 603550 | 10.05916667 | 0.999054374 |
| 0.894444864 | 0.105555205 | 6843.527638 | 0.812162467 | 5.00E-05 | 3155.98995 | 628044 | 10.4674 | 0.99858156 |
| 0.874349802 | 0.125650215 | 7226.582915 | 0.833256115 | 4.00E-05 | 3278.140704 | 652354 | 10.87256667 | 0.999527187 |
| 0.949728548 | 0.050271421 | 7567.170854 | 0.860681659 | 3.00E-05 | 3384.301508 | 673479 | 11.22465 | 0.999054374 |
| 0.884410163 | 0.115589827 | 8543.718593 | 0.861140216 | 2.00E-05 | 3696.969849 | 735703 | 12.26171667 | 0.999054374 |
| 0.859291217 | 0.140708784 | 10002.80905 | 0.868161137 | 1.00E-05 | 4203.839196 | 836569 | 13.94281667 | 0.998108747 |
| 0.849241442 | 0.150758558 | 10314.70854 | 0.856100835 | 9.00E-06 | 4305.170854 | 856730 | 14.27883333 | 0.999054374 |
| 0.84421681 | 0.155783197 | 10521.40704 | 0.856908657 | 8.00E-06 | 4360.839196 | 867813 | 14.46355 | 0.999054374 |
| 0.85929264 | 0.140707354 | 10795.59296 | 0.851536643 | 7.00E-06 | 4624.854271 | 920350 | 15.33916667 | 0.999054374 |
| 0.864318388 | 0.135681609 | 11356.43216 | 0.835860152 | 6.00E-06 | 4864.507538 | 968042 | 16.13403333 | 0.999527187 |
| 0.864318891 | 0.135681109 | 11524.1407 | 0.839217363 | 5.00E-06 | 5046.773869 | 1004311 | 16.73851667 | 0.999527187 |
| 0.81406843 | 0.185931571 | 12755.88945 | 0.82416337 | 4.00E-06 | 5486.366834 | 1091790 | 18.1965 | 0.999054374 |
| 0.83416934 | 0.165830664 | 13319.76884 | 0.84103021 | 3.00E-06 | 5700.80402 | 1134465 | 18.90775 | 0.999054374 |
| 0.76884336 | 0.231156641 | 15362.49749 | 0.857692719 | 2.00E-06 | 6409.427136 | 1275478 | 21.25796667 | 0.999054374 |
| 0.788944292 | 0.211055707 | 17225.44221 | 0.887536976 | 1.00E-06 | 7071.768844 | 1407285 | 23.45475 | 0.999054374 |
| 0.798994592 | 0.201005409 | 18530.14573 | 0.856122219 | 9.00E-07 | 7652.743719 | 1522897 | 25.38161667 | 0.99858156 |
| 0.773869028 | 0.226130972 | 19642.81407 | 0.869472659 | 8.00E-07 | 7851.894472 | 1562527 | 26.04211667 | 0.999527187 |
| 0.773869071 | 0.226130929 | 19718.48241 | 0.872692066 | 7.00E-07 | 8259.356784 | 1643628 | 27.3938 | 0.999527187 |
| 0.798994697 | 0.201005303 | 20614.09548 | 0.873711346 | 6.00E-07 | 8532.537688 | 1697991 | 28.29985 | 0.999527187 |
| 0.743718397 | 0.256281603 | 22309.09548 | 0.861109329 | 5.00E-07 | 8456.236181 | 1682806 | 28.04676667 | 0.999527187 |
| 0.778894301 | 0.221105699 | 22176.38693 | 0.871563491 | 4.00E-07 | 9060.462312 | 1803032 | 30.05053333 | 1 |
| 0.74371848 | 0.25628152 | 25596.88442 | 0.872273899 | 3.00E-07 | 9779.678392 | 1946157 | 32.43595 | 0.999054374 |
| 0.733668267 | 0.266331734 | 31598.69347 | 0.874267318 | 2.00E-07 | 12587.65327 | 2504949 | 41.74915 | 0.999054374 |
| 0.753768805 | 0.246231195 | 37973.84925 | 0.896722383 | 1.00E-07 | 14562.95477 | 2898029 | 48.30048333 | 0.999527187 |
?
?
總結
以上是生活随笔為你收集整理的3层、5层、3层一个卷积核BP神经网络性能比较的全部內容，希望文章能夠幫你解決所遇到的問題。
- 上一篇： ANN神经网络分类2*2矩阵：吸引子和反
- 下一篇： 怎么拆笔记本的cpu风扇 笔记本CPU风