scGAC运行
运行产生的文件
1.预处理：ori_data.tsv产生 data.tsv
2.主函数运行：例如
hidden_Biase.tsv ：图注意自动编码器中的隐藏层，将这个矩阵作为自动化聚类的输入
NE_Biase.csv：网络去噪之后的矩阵
pred_Biase.txt：预测聚类之后的结果
运行
- 查看可选择参数
设置参数为 -h
usage: scGAC.py [-h] [--subtype_path SUBTYPE_PATH] [--k K] [--is_NE IS_NE]
[--PCA_dim PCA_DIM] [--F1 F1] [--F2 F2]
[--n_attn_heads N_ATTN_HEADS] [--dropout_rate DROPOUT_RATE]
[--l2_reg L2_REG] [--learning_rate LEARNING_RATE]
[--pre_lr PRE_LR] [--pre_epochs PRE_EPOCHS] [--epochs EPOCHS]
[--c1 C1] [--c2 C2]
dataset_str n_clusters
positional arguments:
dataset_str name of dataset
n_clusters expected number of clusters
optional arguments:
-h, --help show this help message and exit
--subtype_path SUBTYPE_PATH
path of true labels for evaluation of ARI and NMI
--k K number of neighbors to construct the cell graph
--is_NE IS_NE use NE denoise the cell graph or not
--PCA_dim PCA_DIM dimensionality of input feature matrix that
transformed by PCA
--F1 F1 number of neurons in the 1-st layer of encoder
--F2 F2 number of neurons in the 2-nd layer of encoder
--n_attn_heads N_ATTN_HEADS
number of heads for attention
--dropout_rate DROPOUT_RATE
dropout rate of neurons in autoencoder
--l2_reg L2_REG coefficient for L2 regularizition
--learning_rate LEARNING_RATE
learning rate for training
--pre_lr PRE_LR learning rate for pre-training
--pre_epochs PRE_EPOCHS
number of epochs for pre-training
--epochs EPOCHS number of epochs for pre-training
--c1 C1 weight of reconstruction loss
--c2 C2 weight of clustering loss
Process finished with exit code 0
- Biase 数据集
（1）运行process.py
（2）运行scGAC.py
输入参数 Biase 3（n_clusters）
data/Biase/data.tsv Biase 512 True 3 None
NE
Shape after transformation: (49, 512)
Pre-process: run time is 0.00 minutes
Pre-train: run time is 0.08 minutes
--------------------------------
Kmeans start, with data shape of (49, 64)
Kmeans end
--------------------------------
Iter: 0 , sil_hid: 0.917 , delta_label 0.0 , loss: 0
Iter: 2 , sil_hid: 0.923 , delta_label 0.0 , loss: [1.17 0.5 0.67 0.09]
Iter: 4 , sil_hid: 0.93 , delta_label 0.0 , loss: [ 1.12 0.5 0.61 -0.04]
Iter: 6 , sil_hid: 0.934 , delta_label 0.0 , loss: [1.11 0.51 0.59 0.09]
Iter: 8 , sil_hid: 0.938 , delta_label 0.0 , loss: [1.13 0.51 0.62 0.11]
Iter: 10 , sil_hid: 0.94 , delta_label 0.0 , loss: [ 1.09 0.51 0.58 -0.04]
Iter: 12 , sil_hid: 0.94 , delta_label 0.0 , loss: [ 1.05 0.5 0.56 -0.01]
Iter: 14 , sil_hid: 0.94 , delta_label 0.0 , loss: [ 1.04 0.5 0.54 -0.02]
Iter: 16 , sil_hid: 0.939 , delta_label 0.0 , loss: [1.08 0.51 0.57 0.02]
Iter: 18 , sil_hid: 0.937 , delta_label 0.0 , loss: [ 1.05 0.51 0.54 -0.07]
Iter: 20 , sil_hid: 0.938 , delta_label 0.0 , loss: [ 1.05 0.5 0.55 -0.2 ]
Iter: 22 , sil_hid: 0.94 , delta_label 0.0 , loss: [ 1.02 0.48 0.54 -0.1 ]
Iter: 24 , sil_hid: 0.941 , delta_label 0.0 , loss: [ 1.03 0.49 0.54 -0.1 ]
Iter: 26 , sil_hid: 0.942 , delta_label 0.0 , loss: [ 1.04 0.49 0.55 -0.08]
Iter: 28 , sil_hid: 0.943 , delta_label 0.0 , loss: [ 1.04 0.5 0.54 -0.17]
Iter: 30 , sil_hid: 0.945 , delta_label 0.0 , loss: [ 1.05 0.49 0.55 -0.15]
Iter: 32 , sil_hid: 0.947 , delta_label 0.0 , loss: [ 1. 0.47 0.53 -0.16]
Iter: 34 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 1.03 0.48 0.55 -0.12]
Iter: 36 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 1.03 0.49 0.54 -0.08]
Iter: 38 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 1.01 0.48 0.53 -0.07]
Iter: 40 , sil_hid: 0.947 , delta_label 0.0 , loss: [ 1.04 0.48 0.55 -0.02]
Iter: 42 , sil_hid: 0.947 , delta_label 0.0 , loss: [ 1.04 0.48 0.56 -0.06]
Iter: 44 , sil_hid: 0.946 , delta_label 0.0 , loss: [ 1.04 0.47 0.57 -0.15]
Iter: 46 , sil_hid: 0.943 , delta_label 0.0 , loss: [ 1.02 0.49 0.53 -0.12]
Iter: 48 , sil_hid: 0.941 , delta_label 0.0 , loss: [ 0.97 0.47 0.5 -0.19]
Iter: 50 , sil_hid: 0.94 , delta_label 0.0 , loss: [ 0.98 0.47 0.51 -0.09]
Iter: 52 , sil_hid: 0.941 , delta_label 0.0 , loss: [ 1.06 0.48 0.58 -0.18]
Iter: 54 , sil_hid: 0.942 , delta_label 0.0 , loss: [ 1.01 0.47 0.54 -0.09]
Iter: 56 , sil_hid: 0.944 , delta_label 0.0 , loss: [ 1.01 0.47 0.54 -0.13]
Iter: 58 , sil_hid: 0.946 , delta_label 0.0 , loss: [ 1. 0.47 0.53 -0.11]
Iter: 60 , sil_hid: 0.947 , delta_label 0.0 , loss: [ 1.04 0.47 0.56 -0.1 ]
Iter: 62 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 1.03 0.48 0.55 -0.05]
Iter: 64 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 1.01 0.48 0.53 -0.09]
Iter: 66 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 1.03 0.49 0.54 -0.07]
Iter: 68 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 0.99 0.46 0.53 -0.07]
Iter: 70 , sil_hid: 0.947 , delta_label 0.0 , loss: [ 1. 0.46 0.55 -0.1 ]
Iter: 72 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 1.01 0.47 0.55 -0.07]
Iter: 74 , sil_hid: 0.949 , delta_label 0.0 , loss: [ 0.99 0.46 0.53 -0.11]
Iter: 76 , sil_hid: 0.949 , delta_label 0.0 , loss: [ 1. 0.47 0.53 -0.16]
Iter: 78 , sil_hid: 0.95 , delta_label 0.0 , loss: [ 1. 0.46 0.53 -0.17]
Iter: 80 , sil_hid: 0.95 , delta_label 0.0 , loss: [ 1. 0.46 0.54 -0.14]
Iter: 82 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 0.99 0.47 0.52 -0.14]
Iter: 84 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 1.03 0.48 0.55 -0.19]
Iter: 86 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 0.99 0.46 0.53 -0.11]
Iter: 88 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 1.01 0.48 0.53 -0.08]
Iter: 90 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 1.02 0.47 0.55 -0.11]
Iter: 92 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 1.03 0.47 0.56 -0.11]
Iter: 94 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 1. 0.46 0.54 -0.14]
Iter: 96 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.98 0.46 0.51 -0.1 ]
Iter: 98 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 1. 0.48 0.52 -0.11]
Iter: 100 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.99 0.47 0.52 -0.06]
Iter: 102 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.97 0.46 0.51 -0.12]
Iter: 104 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 0.95 0.47 0.49 -0.18]
Iter: 106 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 0.98 0.47 0.51 -0.16]
Iter: 108 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 0.98 0.47 0.51 -0.1 ]
Iter: 110 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.99 0.46 0.54 -0.2 ]
Iter: 112 , sil_hid: 0.953 , delta_label 0.0 , loss: [ 0.98 0.47 0.5 -0.13]
Iter: 114 , sil_hid: 0.953 , delta_label 0.0 , loss: [ 1. 0.46 0.54 -0.16]
Iter: 116 , sil_hid: 0.954 , delta_label 0.0 , loss: [ 0.98 0.45 0.52 -0.09]
Iter: 118 , sil_hid: 0.954 , delta_label 0.0 , loss: [ 0.99 0.45 0.53 -0.15]
Stop early at 118 epoch
Train: run time is 0.07 minutes
Done.
输入参数 Biase 4（n_clusters）
data/Biase/data.tsv Biase 512 True 4 None
NE
Shape after transformation: (49, 512)
Pre-process: run time is 0.00 minutes
Pre-train: run time is 0.07 minutes
--------------------------------
Kmeans start, with data shape of (49, 64)
Kmeans end
--------------------------------
Iter: 0 , sil_hid: 0.726 , delta_label 0.0 , loss: 0
Iter: 2 , sil_hid: 0.725 , delta_label 0.163 , loss: [1.26 0.52 0.74 0.12]
Iter: 4 , sil_hid: 0.719 , delta_label 0.0 , loss: [1.18 0.52 0.67 0.21]
Iter: 6 , sil_hid: 0.747 , delta_label 0.082 , loss: [1.16 0.51 0.66 0.22]
Iter: 8 , sil_hid: 0.731 , delta_label 0.02 , loss: [1.11 0.51 0.61 0.18]
Iter: 10 , sil_hid: 0.736 , delta_label 0.041 , loss: [1.07 0.52 0.55 0.03]
Iter: 12 , sil_hid: 0.734 , delta_label 0.02 , loss: [ 1.07 0.51 0.55 -0.02]
Iter: 14 , sil_hid: 0.756 , delta_label 0.082 , loss: [ 1.02 0.5 0.52 -0.01]
Iter: 16 , sil_hid: 0.759 , delta_label 0.0 , loss: [1.04 0.5 0.53 0.02]
Iter: 18 , sil_hid: 0.758 , delta_label 0.0 , loss: [1.07 0.5 0.56 0.07]
Iter: 20 , sil_hid: 0.761 , delta_label 0.0 , loss: [ 1.05 0.49 0.56 -0.05]
Iter: 22 , sil_hid: 0.738 , delta_label 0.061 , loss: [ 1.06 0.51 0.55 -0.09]
Iter: 24 , sil_hid: 0.731 , delta_label 0.0 , loss: [1.03 0.5 0.54 0.03]
Iter: 26 , sil_hid: 0.76 , delta_label 0.061 , loss: [ 1.1 0.51 0.59 -0.14]
Iter: 28 , sil_hid: 0.765 , delta_label 0.0 , loss: [ 1.05 0.51 0.54 -0.06]
Iter: 30 , sil_hid: 0.764 , delta_label 0.0 , loss: [ 1. 0.49 0.52 -0.03]
Iter: 32 , sil_hid: 0.758 , delta_label 0.0 , loss: [ 1.02 0.47 0.54 -0.04]
Iter: 34 , sil_hid: 0.752 , delta_label 0.0 , loss: [1.02 0.5 0.52 0.02]
Iter: 36 , sil_hid: 0.723 , delta_label 0.061 , loss: [1.03 0.52 0.51 0.01]
Iter: 38 , sil_hid: 0.735 , delta_label 0.041 , loss: [ 0.99 0.48 0.51 -0.02]
Iter: 40 , sil_hid: 0.739 , delta_label 0.02 , loss: [ 0.97 0.49 0.47 -0. ]
Iter: 42 , sil_hid: 0.721 , delta_label 0.041 , loss: [ 1.01 0.49 0.53 -0.14]
Iter: 44 , sil_hid: 0.72 , delta_label 0.0 , loss: [ 1. 0.49 0.51 -0.08]
Iter: 46 , sil_hid: 0.737 , delta_label 0.041 , loss: [ 0.96 0.48 0.47 -0.06]
Iter: 48 , sil_hid: 0.735 , delta_label 0.02 , loss: [0.98 0.49 0.49 0.01]
Iter: 50 , sil_hid: 0.729 , delta_label 0.02 , loss: [ 0.98 0.48 0.5 -0.1 ]
Iter: 52 , sil_hid: 0.729 , delta_label 0.0 , loss: [ 0.96 0.49 0.47 -0.03]
Iter: 54 , sil_hid: 0.729 , delta_label 0.0 , loss: [ 0.94 0.48 0.46 -0.09]
Iter: 56 , sil_hid: 0.73 , delta_label 0.02 , loss: [ 0.95 0.47 0.48 -0.19]
Iter: 58 , sil_hid: 0.768 , delta_label 0.061 , loss: [ 0.95 0.48 0.48 -0.12]
Iter: 60 , sil_hid: 0.773 , delta_label 0.0 , loss: [ 0.95 0.48 0.48 -0.13]
Iter: 62 , sil_hid: 0.771 , delta_label 0.082 , loss: [ 0.95 0.47 0.48 -0.06]
Iter: 64 , sil_hid: 0.769 , delta_label 0.0 , loss: [0.98 0.48 0.51 0. ]
Iter: 66 , sil_hid: 0.765 , delta_label 0.0 , loss: [ 1.03 0.51 0.53 -0.01]
Iter: 68 , sil_hid: 0.761 , delta_label 0.0 , loss: [1.02 0.48 0.55 0. ]
Iter: 70 , sil_hid: 0.76 , delta_label 0.0 , loss: [ 1.07 0.5 0.57 -0.05]
Iter: 72 , sil_hid: 0.76 , delta_label 0.0 , loss: [ 1.01 0.47 0.53 -0.02]
Iter: 74 , sil_hid: 0.766 , delta_label 0.082 , loss: [ 1.02 0.47 0.55 -0.03]
Iter: 76 , sil_hid: 0.763 , delta_label 0.082 , loss: [ 1.01 0.47 0.54 -0.04]
Iter: 78 , sil_hid: 0.764 , delta_label 0.0 , loss: [ 1.01 0.47 0.54 -0.07]
Iter: 80 , sil_hid: 0.763 , delta_label 0.0 , loss: [ 0.98 0.47 0.51 -0.06]
Iter: 82 , sil_hid: 0.769 , delta_label 0.082 , loss: [ 0.95 0.47 0.48 -0.04]
Iter: 84 , sil_hid: 0.76 , delta_label 0.082 , loss: [ 0.94 0.46 0.48 -0.07]
Iter: 86 , sil_hid: 0.764 , delta_label 0.082 , loss: [ 0.96 0.47 0.49 -0.06]
Iter: 88 , sil_hid: 0.761 , delta_label 0.0 , loss: [ 0.94 0.46 0.48 -0.06]
Iter: 90 , sil_hid: 0.761 , delta_label 0.0 , loss: [ 0.96 0.46 0.5 -0.01]
Iter: 92 , sil_hid: 0.76 , delta_label 0.0 , loss: [ 0.97 0.46 0.51 -0.09]
Iter: 94 , sil_hid: 0.756 , delta_label 0.082 , loss: [ 0.96 0.45 0.5 -0.12]
Iter: 96 , sil_hid: 0.773 , delta_label 0.02 , loss: [ 0.99 0.47 0.52 -0.08]
Iter: 98 , sil_hid: 0.756 , delta_label 0.02 , loss: [ 0.99 0.48 0.51 -0.02]
Iter: 100 , sil_hid: 0.754 , delta_label 0.0 , loss: [ 0.99 0.47 0.51 -0.03]
Iter: 102 , sil_hid: 0.765 , delta_label 0.082 , loss: [ 0.98 0.46 0.52 -0. ]
Iter: 104 , sil_hid: 0.742 , delta_label 0.102 , loss: [1. 0.48 0.53 0.02]
Iter: 106 , sil_hid: 0.739 , delta_label 0.02 , loss: [ 0.97 0.47 0.51 -0.09]
Iter: 108 , sil_hid: 0.738 , delta_label 0.0 , loss: [ 0.96 0.45 0.51 -0.08]
Iter: 110 , sil_hid: 0.739 , delta_label 0.02 , loss: [ 0.96 0.46 0.5 -0.01]
Iter: 112 , sil_hid: 0.733 , delta_label 0.0 , loss: [0.99 0.46 0.53 0.01]
Iter: 114 , sil_hid: 0.735 , delta_label 0.02 , loss: [ 0.96 0.47 0.5 -0.03]
Iter: 116 , sil_hid: 0.736 , delta_label 0.0 , loss: [ 0.99 0.48 0.52 -0.07]
Iter: 118 , sil_hid: 0.739 , delta_label 0.02 , loss: [ 0.97 0.47 0.5 -0.05]
Stop early at 118 epoch
Train: run time is 0.09 minutes
Done.
输入参数 Biase 3 --subtype_path data/Biase/subtype.ann --k 4
data/Biase/data.tsv Biase 512 True 3 4
NE
Shape after transformation: (49, 512)
Pre-process: run time is 0.00 minutes**
Pre-train: run time is 0.07 minutes
--------------------------------
Kmeans start, with data shape of (49, 64)
Kmeans end
--------------------------------
Iter: 0 , sil_hid: 0.922 , delta_label 0.0 , loss: 0
Iter: 2 , sil_hid: 0.924 , delta_label 0.0 , loss: [1.19 0.52 0.67 0.12]
Iter: 4 , sil_hid: 0.928 , delta_label 0.0 , loss: [1.15 0.52 0.63 0.02]
Iter: 6 , sil_hid: 0.932 , delta_label 0.0 , loss: [ 1.13 0.54 0.59 -0.05]
Iter: 8 , sil_hid: 0.936 , delta_label 0.0 , loss: [1.14 0.54 0.6 0.04]
Iter: 10 , sil_hid: 0.939 , delta_label 0.0 , loss: [ 1.08 0.51 0.57 -0.02]
Iter: 12 , sil_hid: 0.94 , delta_label 0.0 , loss: [ 1.11 0.51 0.6 -0. ]
Iter: 14 , sil_hid: 0.942 , delta_label 0.0 , loss: [1.06 0.49 0.56 0.05]
Iter: 16 , sil_hid: 0.943 , delta_label 0.0 , loss: [1.11 0.52 0.59 0.04]
Iter: 18 , sil_hid: 0.943 , delta_label 0.0 , loss: [1.08 0.5 0.58 0.03]
Iter: 20 , sil_hid: 0.943 , delta_label 0.0 , loss: [1.09 0.5 0.59 0.01]
Iter: 22 , sil_hid: 0.942 , delta_label 0.0 , loss: [ 1.05 0.5 0.55 -0.01]
Iter: 24 , sil_hid: 0.941 , delta_label 0.0 , loss: [ 1.06 0.49 0.57 -0.02]
Iter: 26 , sil_hid: 0.941 , delta_label 0.0 , loss: [ 1.04 0.5 0.54 -0.13]
Iter: 28 , sil_hid: 0.941 , delta_label 0.0 , loss: [ 1.01 0.48 0.53 -0.04]
Iter: 30 , sil_hid: 0.94 , delta_label 0.0 , loss: [ 1.08 0.52 0.56 -0.05]
Iter: 32 , sil_hid: 0.94 , delta_label 0.0 , loss: [ 1.04 0.5 0.54 -0.07]
Iter: 34 , sil_hid: 0.941 , delta_label 0.0 , loss: [ 1.02 0.48 0.54 -0.11]
Iter: 36 , sil_hid: 0.941 , delta_label 0.0 , loss: [ 1.03 0.49 0.54 -0.06]
Iter: 38 , sil_hid: 0.941 , delta_label 0.0 , loss: [ 1.02 0.49 0.53 -0.04]
Iter: 40 , sil_hid: 0.942 , delta_label 0.0 , loss: [ 1.02 0.49 0.53 -0.1 ]
Iter: 42 , sil_hid: 0.942 , delta_label 0.0 , loss: [ 1.02 0.48 0.54 -0.03]
Iter: 44 , sil_hid: 0.943 , delta_label 0.0 , loss: [ 1.03 0.5 0.54 -0.14]
Iter: 46 , sil_hid: 0.943 , delta_label 0.0 , loss: [ 0.97 0.47 0.5 -0.06]
Iter: 48 , sil_hid: 0.944 , delta_label 0.0 , loss: [ 1.04 0.5 0.55 -0.13]
Iter: 50 , sil_hid: 0.946 , delta_label 0.0 , loss: [ 1.01 0.49 0.52 -0.09]
Iter: 52 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 0.99 0.47 0.52 -0.12]
Iter: 54 , sil_hid: 0.949 , delta_label 0.0 , loss: [ 0.99 0.47 0.52 -0.14]
Iter: 56 , sil_hid: 0.95 , delta_label 0.0 , loss: [ 1.03 0.48 0.55 -0.1 ]
Iter: 58 , sil_hid: 0.95 , delta_label 0.0 , loss: [ 1.01 0.47 0.54 -0.13]
Iter: 60 , sil_hid: 0.949 , delta_label 0.0 , loss: [ 1.02 0.49 0.53 -0.11]
Iter: 62 , sil_hid: 0.949 , delta_label 0.0 , loss: [ 1.01 0.47 0.54 -0.21]
Iter: 64 , sil_hid: 0.949 , delta_label 0.0 , loss: [ 1.02 0.47 0.55 -0.16]
Iter: 66 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 0.96 0.47 0.5 -0.11]
Iter: 68 , sil_hid: 0.947 , delta_label 0.0 , loss: [ 0.97 0.46 0.5 -0.18]
Iter: 70 , sil_hid: 0.947 , delta_label 0.0 , loss: [ 0.96 0.47 0.49 -0.2 ]
Iter: 72 , sil_hid: 0.947 , delta_label 0.0 , loss: [ 0.99 0.48 0.51 -0.15]
Iter: 74 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 0.98 0.48 0.5 -0.09]
Iter: 76 , sil_hid: 0.948 , delta_label 0.0 , loss: [ 0.96 0.47 0.49 -0.15]
Iter: 78 , sil_hid: 0.949 , delta_label 0.0 , loss: [ 0.97 0.46 0.51 -0.15]
Iter: 80 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 1.01 0.48 0.52 -0.11]
Iter: 82 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.95 0.47 0.48 -0.14]
Iter: 84 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 1.01 0.48 0.53 -0.13]
Iter: 86 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.98 0.48 0.51 -0.05]
Iter: 88 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.99 0.47 0.51 -0.07]
Iter: 90 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.98 0.46 0.52 -0.13]
Iter: 92 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 0.97 0.46 0.51 -0.13]
Iter: 94 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 1.03 0.47 0.55 -0.04]
Iter: 96 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 0.98 0.47 0.51 -0.12]
Iter: 98 , sil_hid: 0.951 , delta_label 0.0 , loss: [ 1. 0.49 0.51 -0.15]
Iter: 100 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.99 0.48 0.51 -0.2 ]
Iter: 102 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 1. 0.47 0.53 -0.09]
Iter: 104 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.99 0.47 0.52 -0.09]
Iter: 106 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 1.01 0.48 0.53 -0.11]
Iter: 108 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.98 0.46 0.52 -0.11]
Iter: 110 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.97 0.47 0.51 -0.1 ]
Iter: 112 , sil_hid: 0.952 , delta_label 0.0 , loss: [ 0.98 0.47 0.52 -0.07]
Iter: 114 , sil_hid: 0.953 , delta_label 0.0 , loss: [ 0.98 0.47 0.51 -0.06]
Iter: 116 , sil_hid: 0.953 , delta_label 0.0 , loss: [ 0.98 0.46 0.52 -0.12]
Iter: 118 , sil_hid: 0.954 , delta_label 0.0 , loss: [ 1. 0.46 0.54 -0.12]
Stop early at 118 epoch
Train: run time is 0.08 minutes
#######################
ARI 1.0
NMI 1.0
Done.
- Björklund数据集
基因名称为第一列
例如
（1）运行process.py
（2）scGAC.py
输入参数：Björklund 3
data/Björklund/data.tsv Björklund 512 True 3 None
NE
Shape after transformation: (648, 512)
Pre-process: run time is 0.10 minutes
Pre-train: run time is 0.60 minutes
--------------------------------
Kmeans start, with data shape of (648, 64)
Kmeans end
--------------------------------
Iter: 0 , sil_hid: 0.609 , delta_label 0.0 , loss: 0
Iter: 2 , sil_hid: 0.639 , delta_label 0.014 , loss: [ 1.36 0.77 0.59 -0.08]
Iter: 4 , sil_hid: 0.66 , delta_label 0.005 , loss: [ 1.39 0.77 0.62 -0.04]
Iter: 6 , sil_hid: 0.674 , delta_label 0.006 , loss: [ 1.32 0.77 0.55 -0.05]
Iter: 8 , sil_hid: 0.682 , delta_label 0.008 , loss: [ 1.3 0.77 0.53 -0.09]
Iter: 10 , sil_hid: 0.688 , delta_label 0.005 , loss: [ 1.25 0.77 0.48 -0.02]
Iter: 12 , sil_hid: 0.695 , delta_label 0.0 , loss: [ 1.24 0.77 0.47 -0.05]
Iter: 14 , sil_hid: 0.702 , delta_label 0.002 , loss: [ 1.21 0.77 0.44 -0.01]
Iter: 16 , sil_hid: 0.708 , delta_label 0.002 , loss: [ 1.18 0.77 0.41 -0.06]
Iter: 18 , sil_hid: 0.712 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.05]
Iter: 20 , sil_hid: 0.716 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.05]
Iter: 22 , sil_hid: 0.719 , delta_label 0.0 , loss: [ 1.14 0.77 0.38 -0.12]
Iter: 24 , sil_hid: 0.721 , delta_label 0.002 , loss: [ 1.15 0.77 0.38 -0.1 ]
Iter: 26 , sil_hid: 0.724 , delta_label 0.002 , loss: [ 1.15 0.77 0.38 -0.09]
Iter: 28 , sil_hid: 0.727 , delta_label 0.0 , loss: [ 1.15 0.77 0.38 -0.16]
Iter: 30 , sil_hid: 0.73 , delta_label 0.002 , loss: [ 1.14 0.77 0.38 -0.09]
Iter: 32 , sil_hid: 0.734 , delta_label 0.002 , loss: [ 1.14 0.77 0.38 -0.1 ]
Iter: 34 , sil_hid: 0.737 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.09]
Iter: 36 , sil_hid: 0.738 , delta_label 0.003 , loss: [ 1.15 0.77 0.38 -0.12]
Iter: 38 , sil_hid: 0.74 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.13]
Iter: 40 , sil_hid: 0.742 , delta_label 0.0 , loss: [ 1.15 0.77 0.38 -0.12]
Iter: 42 , sil_hid: 0.744 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.08]
Iter: 44 , sil_hid: 0.745 , delta_label 0.003 , loss: [ 1.16 0.77 0.39 -0.1 ]
Iter: 46 , sil_hid: 0.746 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.09]
Iter: 48 , sil_hid: 0.749 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.09]
Iter: 50 , sil_hid: 0.75 , delta_label 0.003 , loss: [ 1.16 0.77 0.39 -0.11]
Iter: 52 , sil_hid: 0.751 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.12]
Iter: 54 , sil_hid: 0.753 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.09]
Iter: 56 , sil_hid: 0.755 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.14]
Iter: 58 , sil_hid: 0.757 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.11]
Iter: 60 , sil_hid: 0.758 , delta_label 0.0 , loss: [ 1.16 0.77 0.4 -0.09]
Iter: 62 , sil_hid: 0.759 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.1 ]
Iter: 64 , sil_hid: 0.761 , delta_label 0.0 , loss: [ 1.15 0.77 0.39 -0.13]
Iter: 66 , sil_hid: 0.762 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.12]
Iter: 68 , sil_hid: 0.761 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.12]
Iter: 70 , sil_hid: 0.762 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.14]
Iter: 72 , sil_hid: 0.763 , delta_label 0.003 , loss: [ 1.17 0.77 0.4 -0.08]
Iter: 74 , sil_hid: 0.763 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.07]
Iter: 76 , sil_hid: 0.764 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.08]
Iter: 78 , sil_hid: 0.764 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.1 ]
Iter: 80 , sil_hid: 0.764 , delta_label 0.0 , loss: [ 1.15 0.77 0.38 -0.07]
Iter: 82 , sil_hid: 0.766 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.08]
Iter: 84 , sil_hid: 0.766 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.15]
Iter: 86 , sil_hid: 0.766 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.08]
Iter: 88 , sil_hid: 0.766 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.11]
Iter: 90 , sil_hid: 0.767 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.17]
Iter: 92 , sil_hid: 0.768 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.07]
Iter: 94 , sil_hid: 0.768 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.08]
Iter: 96 , sil_hid: 0.767 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.1 ]
Iter: 98 , sil_hid: 0.768 , delta_label 0.0 , loss: [ 1.17 0.77 0.41 -0.09]
Iter: 100 , sil_hid: 0.768 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.08]
Iter: 102 , sil_hid: 0.77 , delta_label 0.003 , loss: [ 1.17 0.77 0.4 -0.09]
Iter: 104 , sil_hid: 0.77 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.09]
Iter: 106 , sil_hid: 0.771 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.11]
Iter: 108 , sil_hid: 0.772 , delta_label 0.002 , loss: [ 1.16 0.77 0.4 -0.14]
Iter: 110 , sil_hid: 0.771 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.09]
Iter: 112 , sil_hid: 0.771 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.12]
Iter: 114 , sil_hid: 0.771 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.1 ]
Iter: 116 , sil_hid: 0.772 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.08]
Iter: 118 , sil_hid: 0.772 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.1 ]
Iter: 120 , sil_hid: 0.773 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.12]
Iter: 122 , sil_hid: 0.773 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.14]
Iter: 124 , sil_hid: 0.774 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.12]
Iter: 126 , sil_hid: 0.774 , delta_label 0.0 , loss: [ 1.16 0.77 0.4 -0.11]
Iter: 128 , sil_hid: 0.775 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.08]
Iter: 130 , sil_hid: 0.775 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.07]
Iter: 132 , sil_hid: 0.776 , delta_label 0.002 , loss: [ 1.16 0.77 0.39 -0.07]
Iter: 134 , sil_hid: 0.778 , delta_label 0.003 , loss: [ 1.16 0.77 0.39 -0.1 ]
Iter: 136 , sil_hid: 0.781 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.1 ]
Iter: 138 , sil_hid: 0.781 , delta_label 0.0 , loss: [ 1.17 0.77 0.41 -0.11]
Iter: 140 , sil_hid: 0.781 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.1 ]
Iter: 142 , sil_hid: 0.782 , delta_label 0.0 , loss: [ 1.18 0.77 0.42 -0.11]
Iter: 144 , sil_hid: 0.783 , delta_label 0.002 , loss: [ 1.18 0.77 0.41 -0.09]
Iter: 146 , sil_hid: 0.784 , delta_label 0.006 , loss: [ 1.18 0.77 0.41 -0.07]
Iter: 148 , sil_hid: 0.785 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.1 ]
Iter: 150 , sil_hid: 0.786 , delta_label 0.0 , loss: [ 1.17 0.77 0.41 -0.11]
Stop early at 150 epoch
Train: run time is 0.79 minutes
Done.
输入参数 Björklund 3 --k 4
data/Björklund/data.tsv Björklund 512 True 3 4
NE
Shape after transformation: (648, 512)
Pre-process: run time is 0.11 minutes
Pre-train: run time is 0.50 minutes
--------------------------------
Kmeans start, with data shape of (648, 64)
Kmeans end
--------------------------------
Iter: 0 , sil_hid: 0.67 , delta_label 0.0 , loss: 0
Iter: 2 , sil_hid: 0.686 , delta_label 0.002 , loss: [ 1.39 0.77 0.62 -0.13]
Iter: 4 , sil_hid: 0.701 , delta_label 0.003 , loss: [ 1.35 0.77 0.59 -0.1 ]
Iter: 6 , sil_hid: 0.711 , delta_label 0.008 , loss: [ 1.35 0.77 0.58 -0.09]
Iter: 8 , sil_hid: 0.72 , delta_label 0.005 , loss: [ 1.32 0.77 0.55 -0.15]
Iter: 10 , sil_hid: 0.726 , delta_label 0.002 , loss: [ 1.29 0.77 0.52 -0.14]
Iter: 12 , sil_hid: 0.731 , delta_label 0.002 , loss: [ 1.25 0.77 0.48 -0.11]
Iter: 14 , sil_hid: 0.736 , delta_label 0.0 , loss: [ 1.23 0.77 0.46 -0.12]
Iter: 16 , sil_hid: 0.741 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.16]
Iter: 18 , sil_hid: 0.745 , delta_label 0.0 , loss: [ 1.19 0.77 0.43 -0.09]
Iter: 20 , sil_hid: 0.748 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.1 ]
Iter: 22 , sil_hid: 0.753 , delta_label 0.002 , loss: [ 1.19 0.77 0.42 -0.15]
Iter: 24 , sil_hid: 0.759 , delta_label 0.002 , loss: [ 1.21 0.77 0.44 -0.09]
Iter: 26 , sil_hid: 0.764 , delta_label 0.002 , loss: [ 1.2 0.77 0.43 -0.18]
Iter: 28 , sil_hid: 0.768 , delta_label 0.002 , loss: [ 1.2 0.77 0.43 -0.14]
Iter: 30 , sil_hid: 0.771 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.13]
Iter: 32 , sil_hid: 0.775 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.15]
Iter: 34 , sil_hid: 0.778 , delta_label 0.002 , loss: [ 1.22 0.77 0.45 -0.16]
Iter: 36 , sil_hid: 0.783 , delta_label 0.002 , loss: [ 1.21 0.77 0.45 -0.14]
Iter: 38 , sil_hid: 0.786 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.17]
Iter: 40 , sil_hid: 0.789 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.12]
Iter: 42 , sil_hid: 0.791 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.14]
Iter: 44 , sil_hid: 0.793 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.15]
Iter: 46 , sil_hid: 0.796 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.13]
Iter: 48 , sil_hid: 0.797 , delta_label 0.0 , loss: [ 1.21 0.77 0.45 -0.14]
Iter: 50 , sil_hid: 0.798 , delta_label 0.003 , loss: [ 1.21 0.77 0.45 -0.18]
Iter: 52 , sil_hid: 0.799 , delta_label 0.002 , loss: [ 1.22 0.77 0.45 -0.14]
Iter: 54 , sil_hid: 0.799 , delta_label 0.002 , loss: [ 1.21 0.77 0.44 -0.15]
Iter: 56 , sil_hid: 0.8 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.13]
Iter: 58 , sil_hid: 0.8 , delta_label 0.0 , loss: [ 1.21 0.77 0.45 -0.17]
Iter: 60 , sil_hid: 0.801 , delta_label 0.002 , loss: [ 1.22 0.77 0.45 -0.12]
Iter: 62 , sil_hid: 0.802 , delta_label 0.0 , loss: [ 1.22 0.77 0.45 -0.2 ]
Iter: 64 , sil_hid: 0.802 , delta_label 0.0 , loss: [ 1.22 0.77 0.45 -0.16]
Iter: 66 , sil_hid: 0.803 , delta_label 0.002 , loss: [ 1.22 0.77 0.45 -0.17]
Iter: 68 , sil_hid: 0.803 , delta_label 0.0 , loss: [ 1.21 0.77 0.45 -0.19]
Iter: 70 , sil_hid: 0.802 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.13]
Iter: 72 , sil_hid: 0.802 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.12]
Iter: 74 , sil_hid: 0.802 , delta_label 0.0 , loss: [ 1.22 0.77 0.45 -0.14]
Iter: 76 , sil_hid: 0.802 , delta_label 0.0 , loss: [ 1.2 0.77 0.44 -0.17]
Iter: 78 , sil_hid: 0.802 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.15]
Iter: 80 , sil_hid: 0.802 , delta_label 0.0 , loss: [ 1.2 0.77 0.44 -0.16]
Iter: 82 , sil_hid: 0.803 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.14]
Iter: 84 , sil_hid: 0.804 , delta_label 0.002 , loss: [ 1.22 0.77 0.45 -0.13]
Iter: 86 , sil_hid: 0.805 , delta_label 0.0 , loss: [ 1.22 0.77 0.45 -0.15]
Iter: 88 , sil_hid: 0.805 , delta_label 0.0 , loss: [ 1.21 0.77 0.45 -0.14]
Iter: 90 , sil_hid: 0.806 , delta_label 0.002 , loss: [ 1.23 0.77 0.46 -0.2 ]
Iter: 92 , sil_hid: 0.806 , delta_label 0.0 , loss: [ 1.22 0.77 0.45 -0.15]
Iter: 94 , sil_hid: 0.806 , delta_label 0.002 , loss: [ 1.22 0.77 0.45 -0.13]
Iter: 96 , sil_hid: 0.806 , delta_label 0.003 , loss: [ 1.23 0.77 0.46 -0.13]
Iter: 98 , sil_hid: 0.807 , delta_label 0.0 , loss: [ 1.22 0.77 0.46 -0.17]
Iter: 100 , sil_hid: 0.807 , delta_label 0.0 , loss: [ 1.23 0.77 0.46 -0.2 ]
Iter: 102 , sil_hid: 0.807 , delta_label 0.002 , loss: [ 1.23 0.77 0.46 -0.16]
Iter: 104 , sil_hid: 0.808 , delta_label 0.0 , loss: [ 1.22 0.77 0.45 -0.14]
Iter: 106 , sil_hid: 0.808 , delta_label 0.0 , loss: [ 1.22 0.77 0.46 -0.14]
Iter: 108 , sil_hid: 0.81 , delta_label 0.002 , loss: [ 1.21 0.77 0.44 -0.13]
Iter: 110 , sil_hid: 0.811 , delta_label 0.0 , loss: [ 1.23 0.77 0.46 -0.15]
Iter: 112 , sil_hid: 0.811 , delta_label 0.0 , loss: [ 1.22 0.77 0.46 -0.15]
Iter: 114 , sil_hid: 0.812 , delta_label 0.0 , loss: [ 1.23 0.77 0.46 -0.16]
Iter: 116 , sil_hid: 0.812 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.11]
Iter: 118 , sil_hid: 0.812 , delta_label 0.0 , loss: [ 1.22 0.77 0.45 -0.14]
Iter: 120 , sil_hid: 0.813 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.16]
Iter: 122 , sil_hid: 0.815 , delta_label 0.002 , loss: [ 1.22 0.77 0.45 -0.16]
Iter: 124 , sil_hid: 0.816 , delta_label 0.0 , loss: [ 1.22 0.77 0.46 -0.1 ]
Iter: 126 , sil_hid: 0.817 , delta_label 0.0 , loss: [ 1.22 0.77 0.46 -0.14]
Iter: 128 , sil_hid: 0.817 , delta_label 0.002 , loss: [ 1.22 0.77 0.45 -0.12]
Iter: 130 , sil_hid: 0.817 , delta_label 0.0 , loss: [ 1.21 0.77 0.45 -0.13]
Iter: 132 , sil_hid: 0.817 , delta_label 0.003 , loss: [ 1.22 0.77 0.46 -0.16]
Iter: 134 , sil_hid: 0.817 , delta_label 0.0 , loss: [ 1.21 0.77 0.45 -0.12]
Iter: 136 , sil_hid: 0.817 , delta_label 0.0 , loss: [ 1.23 0.77 0.46 -0.18]
Iter: 138 , sil_hid: 0.817 , delta_label 0.0 , loss: [ 1.22 0.77 0.45 -0.18]
Iter: 140 , sil_hid: 0.817 , delta_label 0.0 , loss: [ 1.21 0.77 0.45 -0.15]
Iter: 142 , sil_hid: 0.817 , delta_label 0.0 , loss: [ 1.21 0.77 0.45 -0.17]
Stop early at 142 epoch
Train: run time is 0.41 minutes
Done.
基因名称为第二列
例如
输入数据名称 3
data/Björklund/data.tsv Björklund 512 True 3 None
NE
Shape after transformation: (648, 512)
Pre-process: run time is 0.22 minutes
Pre-train: run time is 0.44 minutes
--------------------------------
Kmeans start, with data shape of (648, 64)
Kmeans end
--------------------------------
Iter: 0 , sil_hid: 0.646 , delta_label 0.0 , loss: 0
Iter: 2 , sil_hid: 0.671 , delta_label 0.012 , loss: [ 1.4 0.77 0.64 -0.09]
Iter: 4 , sil_hid: 0.69 , delta_label 0.008 , loss: [ 1.39 0.77 0.62 -0.1 ]
Iter: 6 , sil_hid: 0.705 , delta_label 0.005 , loss: [ 1.34 0.77 0.57 -0.1 ]
Iter: 8 , sil_hid: 0.713 , delta_label 0.005 , loss: [ 1.28 0.77 0.51 -0.12]
Iter: 10 , sil_hid: 0.72 , delta_label 0.003 , loss: [ 1.25 0.77 0.48 -0.09]
Iter: 12 , sil_hid: 0.726 , delta_label 0.0 , loss: [ 1.23 0.77 0.46 -0.07]
Iter: 14 , sil_hid: 0.731 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.06]
Iter: 16 , sil_hid: 0.734 , delta_label 0.0 , loss: [ 1.19 0.77 0.42 -0.1 ]
Iter: 18 , sil_hid: 0.736 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.1 ]
Iter: 20 , sil_hid: 0.74 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.08]
Iter: 22 , sil_hid: 0.742 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.13]
Iter: 24 , sil_hid: 0.744 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.12]
Iter: 26 , sil_hid: 0.747 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.12]
Iter: 28 , sil_hid: 0.748 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.13]
Iter: 30 , sil_hid: 0.749 , delta_label 0.0 , loss: [ 1.16 0.77 0.39 -0.1 ]
Iter: 32 , sil_hid: 0.751 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.15]
Iter: 34 , sil_hid: 0.751 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.15]
Iter: 36 , sil_hid: 0.752 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.15]
Iter: 38 , sil_hid: 0.752 , delta_label 0.0 , loss: [ 1.19 0.77 0.42 -0.18]
Iter: 40 , sil_hid: 0.754 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.13]
Iter: 42 , sil_hid: 0.755 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.11]
Iter: 44 , sil_hid: 0.757 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.16]
Iter: 46 , sil_hid: 0.76 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.13]
Iter: 48 , sil_hid: 0.763 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.15]
Iter: 50 , sil_hid: 0.765 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.2 ]
Iter: 52 , sil_hid: 0.766 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.15]
Iter: 54 , sil_hid: 0.767 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.15]
Iter: 56 , sil_hid: 0.768 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.13]
Iter: 58 , sil_hid: 0.77 , delta_label 0.002 , loss: [ 1.18 0.77 0.41 -0.15]
Iter: 60 , sil_hid: 0.771 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.17]
Iter: 62 , sil_hid: 0.772 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.12]
Iter: 64 , sil_hid: 0.773 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.2 ]
Iter: 66 , sil_hid: 0.775 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.15]
Iter: 68 , sil_hid: 0.776 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.19]
Iter: 70 , sil_hid: 0.777 , delta_label 0.0 , loss: [ 1.19 0.77 0.43 -0.22]
Iter: 72 , sil_hid: 0.778 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.15]
Iter: 74 , sil_hid: 0.779 , delta_label 0.002 , loss: [ 1.18 0.77 0.41 -0.19]
Iter: 76 , sil_hid: 0.78 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.18]
Iter: 78 , sil_hid: 0.782 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.14]
Iter: 80 , sil_hid: 0.784 , delta_label 0.002 , loss: [ 1.18 0.77 0.42 -0.18]
Iter: 82 , sil_hid: 0.785 , delta_label 0.0 , loss: [ 1.19 0.77 0.42 -0.11]
Iter: 84 , sil_hid: 0.785 , delta_label 0.0 , loss: [ 1.19 0.77 0.42 -0.1 ]
Iter: 86 , sil_hid: 0.785 , delta_label 0.0 , loss: [ 1.18 0.77 0.42 -0.16]
Iter: 88 , sil_hid: 0.786 , delta_label 0.0 , loss: [ 1.19 0.77 0.42 -0.18]
Iter: 90 , sil_hid: 0.787 , delta_label 0.002 , loss: [ 1.19 0.77 0.42 -0.17]
Iter: 92 , sil_hid: 0.788 , delta_label 0.0 , loss: [ 1.19 0.77 0.42 -0.15]
Iter: 94 , sil_hid: 0.789 , delta_label 0.0 , loss: [ 1.19 0.77 0.42 -0.16]
Iter: 96 , sil_hid: 0.79 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.13]
Iter: 98 , sil_hid: 0.792 , delta_label 0.0 , loss: [ 1.19 0.77 0.43 -0.13]
Iter: 100 , sil_hid: 0.793 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.17]
Iter: 102 , sil_hid: 0.792 , delta_label 0.002 , loss: [ 1.19 0.77 0.43 -0.18]
Iter: 104 , sil_hid: 0.793 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.16]
Iter: 106 , sil_hid: 0.792 , delta_label 0.002 , loss: [ 1.19 0.77 0.42 -0.14]
Iter: 108 , sil_hid: 0.792 , delta_label 0.0 , loss: [ 1.19 0.77 0.43 -0.12]
Iter: 110 , sil_hid: 0.793 , delta_label 0.0 , loss: [ 1.18 0.77 0.42 -0.15]
Iter: 112 , sil_hid: 0.793 , delta_label 0.0 , loss: [ 1.18 0.77 0.41 -0.18]
Iter: 114 , sil_hid: 0.794 , delta_label 0.0 , loss: [ 1.17 0.77 0.41 -0.17]
Iter: 116 , sil_hid: 0.794 , delta_label 0.0 , loss: [ 1.17 0.77 0.4 -0.18]
Iter: 118 , sil_hid: 0.794 , delta_label 0.0 , loss: [ 1.17 0.77 0.41 -0.14]
Iter: 120 , sil_hid: 0.796 , delta_label 0.002 , loss: [ 1.17 0.77 0.4 -0.17]
Iter: 122 , sil_hid: 0.796 , delta_label 0.0 , loss: [ 1.17 0.77 0.41 -0.14]
Iter: 124 , sil_hid: 0.797 , delta_label 0.0 , loss: [ 1.17 0.77 0.41 -0.15]
Iter: 126 , sil_hid: 0.797 , delta_label 0.0 , loss: [ 1.17 0.77 0.41 -0.15]
Iter: 128 , sil_hid: 0.797 , delta_label 0.0 , loss: [ 1.18 0.77 0.42 -0.17]
Iter: 130 , sil_hid: 0.799 , delta_label 0.0 , loss: [ 1.18 0.77 0.42 -0.15]
Iter: 132 , sil_hid: 0.8 , delta_label 0.0 , loss: [ 1.19 0.77 0.43 -0.17]
Iter: 134 , sil_hid: 0.802 , delta_label 0.002 , loss: [ 1.19 0.77 0.42 -0.15]
Iter: 136 , sil_hid: 0.803 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.21]
Iter: 138 , sil_hid: 0.804 , delta_label 0.0 , loss: [ 1.19 0.77 0.43 -0.13]
Iter: 140 , sil_hid: 0.804 , delta_label 0.0 , loss: [ 1.19 0.77 0.42 -0.13]
Iter: 142 , sil_hid: 0.805 , delta_label 0.0 , loss: [ 1.19 0.77 0.43 -0.13]
Iter: 144 , sil_hid: 0.805 , delta_label 0.002 , loss: [ 1.21 0.77 0.44 -0.16]
Iter: 146 , sil_hid: 0.806 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.18]
Iter: 148 , sil_hid: 0.806 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.17]
Iter: 150 , sil_hid: 0.807 , delta_label 0.002 , loss: [ 1.21 0.77 0.44 -0.18]
Iter: 152 , sil_hid: 0.808 , delta_label 0.0 , loss: [ 1.21 0.77 0.45 -0.2 ]
Iter: 154 , sil_hid: 0.809 , delta_label 0.0 , loss: [ 1.2 0.77 0.44 -0.12]
Iter: 156 , sil_hid: 0.809 , delta_label 0.0 , loss: [ 1.2 0.77 0.44 -0.16]
Iter: 158 , sil_hid: 0.809 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.11]
Iter: 160 , sil_hid: 0.808 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.14]
Iter: 162 , sil_hid: 0.808 , delta_label 0.0 , loss: [ 1.2 0.77 0.44 -0.15]
Iter: 164 , sil_hid: 0.808 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.1 ]
Iter: 166 , sil_hid: 0.808 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.15]
Iter: 168 , sil_hid: 0.809 , delta_label 0.0 , loss: [ 1.2 0.77 0.43 -0.16]
Iter: 170 , sil_hid: 0.809 , delta_label 0.0 , loss: [ 1.21 0.77 0.44 -0.13]
Iter: 172 , sil_hid: 0.809 , delta_label 0.0 , loss: [ 1.2 0.77 0.44 -0.12]
Iter: 174 , sil_hid: 0.808 , delta_label 0.003 , loss: [ 1.19 0.77 0.42 -0.15]
Iter: 176 , sil_hid: 0.807 , delta_label 0.002 , loss: [ 1.21 0.77 0.45 -0.13]
Stop early at 176 epoch
Train: run time is 0.84 minutes
Done.
Process finished with exit code 0
输入数据名称 4
data/Björklund/data.tsv Björklund 512 True 4 None
NE
Shape after transformation: (648, 512)
Pre-process: run time is 0.22 minutes
Pre-train: run time is 0.75 minutes
--------------------------------
Kmeans start, with data shape of (648, 64)
Kmeans end
--------------------------------
Iter: 0 , sil_hid: 0.578 , delta_label 0.0 , loss: 0
Iter: 2 , sil_hid: 0.592 , delta_label 0.011 , loss: [ 1.54 0.77 0.77 -0.12]
Iter: 4 , sil_hid: 0.602 , delta_label 0.0 , loss: [ 1.52 0.77 0.75 -0.09]
Iter: 6 , sil_hid: 0.602 , delta_label 0.019 , loss: [ 1.46 0.77 0.69 -0.1 ]
Iter: 8 , sil_hid: 0.603 , delta_label 0.011 , loss: [ 1.4 0.77 0.63 -0.12]
Iter: 10 , sil_hid: 0.603 , delta_label 0.008 , loss: [ 1.34 0.77 0.57 -0.13]
Iter: 12 , sil_hid: 0.593 , delta_label 0.015 , loss: [ 1.3 0.77 0.53 -0.13]
Iter: 14 , sil_hid: 0.585 , delta_label 0.012 , loss: [ 1.27 0.77 0.5 -0.12]
Iter: 16 , sil_hid: 0.581 , delta_label 0.008 , loss: [ 1.23 0.77 0.46 -0.14]
Iter: 18 , sil_hid: 0.568 , delta_label 0.015 , loss: [ 1.2 0.77 0.43 -0.12]
Iter: 20 , sil_hid: 0.564 , delta_label 0.009 , loss: [ 1.19 0.77 0.41 -0.12]
Iter: 22 , sil_hid: 0.558 , delta_label 0.011 , loss: [ 1.17 0.77 0.4 -0.1 ]
Iter: 24 , sil_hid: 0.556 , delta_label 0.005 , loss: [ 1.16 0.77 0.39 -0.14]
Iter: 26 , sil_hid: 0.555 , delta_label 0.005 , loss: [ 1.15 0.77 0.38 -0.14]
Iter: 28 , sil_hid: 0.551 , delta_label 0.006 , loss: [ 1.15 0.77 0.38 -0.14]
Iter: 30 , sil_hid: 0.548 , delta_label 0.005 , loss: [ 1.14 0.77 0.37 -0.17]
Iter: 32 , sil_hid: 0.536 , delta_label 0.017 , loss: [ 1.15 0.77 0.38 -0.12]
Iter: 34 , sil_hid: 0.532 , delta_label 0.008 , loss: [ 1.14 0.77 0.37 -0.11]
Iter: 36 , sil_hid: 0.525 , delta_label 0.011 , loss: [ 1.14 0.77 0.37 -0.18]
Iter: 38 , sil_hid: 0.523 , delta_label 0.006 , loss: [ 1.14 0.77 0.37 -0.15]
Iter: 40 , sil_hid: 0.521 , delta_label 0.015 , loss: [ 1.14 0.77 0.37 -0.18]
Iter: 42 , sil_hid: 0.522 , delta_label 0.008 , loss: [ 1.15 0.77 0.38 -0.19]
Iter: 44 , sil_hid: 0.524 , delta_label 0.005 , loss: [ 1.14 0.77 0.37 -0.19]
Iter: 46 , sil_hid: 0.531 , delta_label 0.008 , loss: [ 1.14 0.77 0.38 -0.16]
Iter: 48 , sil_hid: 0.535 , delta_label 0.002 , loss: [ 1.14 0.77 0.37 -0.18]
Iter: 50 , sil_hid: 0.537 , delta_label 0.0 , loss: [ 1.15 0.77 0.38 -0.21]
Iter: 52 , sil_hid: 0.538 , delta_label 0.0 , loss: [ 1.15 0.77 0.38 -0.15]
Iter: 54 , sil_hid: 0.544 , delta_label 0.003 , loss: [ 1.15 0.77 0.38 -0.17]
Iter: 56 , sil_hid: 0.547 , delta_label 0.002 , loss: [ 1.15 0.77 0.38 -0.18]
Iter: 58 , sil_hid: 0.539 , delta_label 0.003 , loss: [ 1.15 0.77 0.38 -0.1 ]
Iter: 60 , sil_hid: 0.55 , delta_label 0.006 , loss: [ 1.15 0.77 0.38 -0.16]
Iter: 62 , sil_hid: 0.559 , delta_label 0.006 , loss: [ 1.15 0.77 0.38 -0.18]
Iter: 64 , sil_hid: 0.566 , delta_label 0.002 , loss: [ 1.14 0.77 0.37 -0.13]
Iter: 66 , sil_hid: 0.589 , delta_label 0.005 , loss: [ 1.16 0.77 0.39 -0.21]
Iter: 68 , sil_hid: 0.612 , delta_label 0.002 , loss: [ 1.15 0.77 0.38 -0.18]
Iter: 70 , sil_hid: 0.617 , delta_label 0.002 , loss: [ 1.15 0.77 0.38 -0.16]
Iter: 72 , sil_hid: 0.621 , delta_label 0.0 , loss: [ 1.14 0.77 0.37 -0.17]
Iter: 74 , sil_hid: 0.625 , delta_label 0.002 , loss: [ 1.15 0.77 0.38 -0.18]
Iter: 76 , sil_hid: 0.568 , delta_label 0.002 , loss: [ 1.15 0.77 0.38 -0.14]
Stop early at 76 epoch
Train: run time is 0.46 minutes
Done.
Process finished with exit code 0