周工作交流

原理

第一步：UTL化，取中间列或者中间行，分别比较相似性（设置阈值），相似性多的中间行或中间列取为标准，将相似的取为另一指标进行覆盖。

第二步：把覆盖的剔除，重复UTL化，接着取

第三步：如此迭代下去，如果得到的patterns不能让整个矩阵的值变得更小，那么就需要进行weak signal detection，来发现patterns。分别从行与列入手，看谁误差小。以列举例，找出两个列和最大的列，让它们做与运算，则得出一个新的列d，用这个d来去expansion，从而得到e。

MEBF

Bidirectional Growth

Weak Signal Detection Algorithm

ART

epoch: 0, Train Loss: 0.061851,Eval Loss: 0.063152, Eval Acc: 0.936848, Eval Pre: nan,Eval Recall: 0.000000
epoch: 1, Train Loss: 0.058144,Eval Loss: 0.063152, Eval Acc: 0.936848, Eval Pre: nan,Eval Recall: 0.000000
epoch: 2, Train Loss: 0.054543,Eval Loss: 0.060646, Eval Acc: 0.939354, Eval Pre: nan,Eval Recall: 0.048654
epoch: 3, Train Loss: 0.050643,Eval Loss: 0.057705, Eval Acc: 0.942295, Eval Pre: nan,Eval Recall: 0.117389
epoch: 4, Train Loss: 0.047969,Eval Loss: 0.055941, Eval Acc: 0.944059, Eval Pre: 0.761611,Eval Recall: 0.167171
epoch: 5, Train Loss: 0.045933,Eval Loss: 0.055455, Eval Acc: 0.944545, Eval Pre: 0.774186,Eval Recall: 0.175179
epoch: 6, Train Loss: 0.044225,Eval Loss: 0.055097, Eval Acc: 0.944903, Eval Pre: 0.737402,Eval Recall: 0.200606
epoch: 7, Train Loss: 0.042716,Eval Loss: 0.054547, Eval Acc: 0.945453, Eval Pre: 0.736079,Eval Recall: 0.215758
epoch: 8, Train Loss: 0.041386,Eval Loss: 0.054751, Eval Acc: 0.945249, Eval Pre: 0.723701,Eval Recall: 0.218687
epoch: 9, Train Loss: 0.040123,Eval Loss: 0.054713, Eval Acc: 0.945287, Eval Pre: 0.714139,Eval Recall: 0.226296
epoch: 10, Train Loss: 0.038899,Eval Loss: 0.054764, Eval Acc: 0.945236, Eval Pre: 0.703424,Eval Recall: 0.233696
epoch: 11, Train Loss: 0.037691,Eval Loss: 0.054611, Eval Acc: 0.945389, Eval Pre: 0.701780,Eval Recall: 0.237202
epoch: 12, Train Loss: 0.036614,Eval Loss: 0.054713, Eval Acc: 0.945287, Eval Pre: 0.691630,Eval Recall: 0.246191
epoch: 13, Train Loss: 0.035457,Eval Loss: 0.054905, Eval Acc: 0.945095, Eval Pre: 0.690341,Eval Recall: 0.241335
epoch: 14, Train Loss: 0.034379,Eval Loss: 0.055250, Eval Acc: 0.944750, Eval Pre: 0.676029,Eval Recall: 0.245459
epoch: 15, Train Loss: 0.033309,Eval Loss: 0.055225, Eval Acc: 0.944775, Eval Pre: 0.663067,Eval Recall: 0.261256
epoch: 16, Train Loss: 0.032288,Eval Loss: 0.055314, Eval Acc: 0.944686, Eval Pre: 0.664900,Eval Recall: 0.256600
epoch: 17, Train Loss: 0.031271,Eval Loss: 0.055263, Eval Acc: 0.944737, Eval Pre: 0.662654,Eval Recall: 0.262584
epoch: 18, Train Loss: 0.030230,Eval Loss: 0.055301, Eval Acc: 0.944699, Eval Pre: 0.662032,Eval Recall: 0.261513
epoch: 19, Train Loss: 0.029217,Eval Loss: 0.056017, Eval Acc: 0.943983, Eval Pre: 0.634722,Eval Recall: 0.276252
epoch: 20, Train Loss: 0.028244,Eval Loss: 0.055723, Eval Acc: 0.944277, Eval Pre: 0.643446,Eval Recall: 0.272547
epoch: 21, Train Loss: 0.027300,Eval Loss: 0.056196, Eval Acc: 0.943804, Eval Pre: 0.634164,Eval Recall: 0.274652
epoch: 22, Train Loss: 0.026335,Eval Loss: 0.056247, Eval Acc: 0.943753, Eval Pre: 0.629612,Eval Recall: 0.276027
epoch: 23, Train Loss: 0.025405,Eval Loss: 0.056810, Eval Acc: 0.943190, Eval Pre: 0.615111,Eval Recall: 0.282923
epoch: 24, Train Loss: 0.024488,Eval Loss: 0.056733, Eval Acc: 0.943267, Eval Pre: 0.616486,Eval Recall: 0.281782
epoch: 25, Train Loss: 0.023583,Eval Loss: 0.056925, Eval Acc: 0.943075, Eval Pre: 0.613671,Eval Recall: 0.280210
epoch: 26, Train Loss: 0.022721,Eval Loss: 0.056797, Eval Acc: 0.943203, Eval Pre: 0.615189,Eval Recall: 0.282543
epoch: 27, Train Loss: 0.021839,Eval Loss: 0.057206, Eval Acc: 0.942794, Eval Pre: 0.605251,Eval Recall: 0.286072
epoch: 28, Train Loss: 0.021032,Eval Loss: 0.057309, Eval Acc: 0.942691, Eval Pre: 0.602579,Eval Recall: 0.286384
epoch: 29, Train Loss: 0.020275,Eval Loss: 0.057718, Eval Acc: 0.942282, Eval Pre: 0.592728,Eval Recall: 0.287771
epoch: 30, Train Loss: 0.019507,Eval Loss: 0.057974, Eval Acc: 0.942026, Eval Pre: 0.584318,Eval Recall: 0.293699
epoch: 31, Train Loss: 0.018760,Eval Loss: 0.058101, Eval Acc: 0.941899, Eval Pre: 0.584987,Eval Recall: 0.289282
epoch: 32, Train Loss: 0.018055,Eval Loss: 0.058204, Eval Acc: 0.941796, Eval Pre: 0.583817,Eval Recall: 0.288418
epoch: 33, Train Loss: 0.017355,Eval Loss: 0.057922, Eval Acc: 0.942078, Eval Pre: 0.588069,Eval Recall: 0.291899
epoch: 34, Train Loss: 0.016679,Eval Loss: 0.058165, Eval Acc: 0.941835, Eval Pre: 0.585181,Eval Recall: 0.285932
epoch: 35, Train Loss: 0.016051,Eval Loss: 0.058357, Eval Acc: 0.941643, Eval Pre: 0.577433,Eval Recall: 0.293414
epoch: 36, Train Loss: 0.015422,Eval Loss: 0.058587, Eval Acc: 0.941413, Eval Pre: 0.575256,Eval Recall: 0.295631
epoch: 37, Train Loss: 0.014833,Eval Loss: 0.058843, Eval Acc: 0.941157, Eval Pre: 0.569486,Eval Recall: 0.298378
epoch: 38, Train Loss: 0.014221,Eval Loss: 0.058587, Eval Acc: 0.941413, Eval Pre: 0.575128,Eval Recall: 0.292361
epoch: 39, Train Loss: 0.013674,Eval Loss: 0.058920, Eval Acc: 0.941080, Eval Pre: 0.570719,Eval Recall: 0.293320
epoch: 40, Train Loss: 0.013134,Eval Loss: 0.059252, Eval Acc: 0.940748, Eval Pre: 0.563740,Eval Recall: 0.294594
epoch: 41, Train Loss: 0.012625,Eval Loss: 0.058856, Eval Acc: 0.941144, Eval Pre: 0.571070,Eval Recall: 0.296034
epoch: 42, Train Loss: 0.012150,Eval Loss: 0.059495, Eval Acc: 0.940505, Eval Pre: 0.556237,Eval Recall: 0.303902
epoch: 43, Train Loss: 0.011674,Eval Loss: 0.059508, Eval Acc: 0.940492, Eval Pre: 0.557009,Eval Recall: 0.304557
epoch: 44, Train Loss: 0.011212,Eval Loss: 0.059201, Eval Acc: 0.940799, Eval Pre: 0.564765,Eval Recall: 0.300135
epoch: 45, Train Loss: 0.010781,Eval Loss: 0.059840, Eval Acc: 0.940160, Eval Pre: 0.552108,Eval Recall: 0.301805
epoch: 46, Train Loss: 0.010367,Eval Loss: 0.059930, Eval Acc: 0.940070, Eval Pre: 0.552668,Eval Recall: 0.301376
epoch: 47, Train Loss: 0.009940,Eval Loss: 0.059968, Eval Acc: 0.940032, Eval Pre: 0.549732,Eval Recall: 0.302651
epoch: 48, Train Loss: 0.009574,Eval Loss: 0.060493, Eval Acc: 0.939507, Eval Pre: 0.542295,Eval Recall: 0.304724
epoch: 49, Train Loss: 0.009189,Eval Loss: 0.060237, Eval Acc: 0.939763, Eval Pre: 0.546306,Eval Recall: 0.300866