support federated linear regression (#583)

* move paillier & ckks into crypto folder

* add regression dataset for VFL

* support federated linear regression

config/node1.yaml

@@ -103,6 +103,12 @@ datasets:
     source: "data/FL/multiclass/hfl/test/client1.csv"
 
   # FL regression
+  - description: "regression_vfl_train_host"
+    model: "csv"
+    source: "data/FL/regression/vfl/train/host.csv"
+  - description: "regression_vfl_test_host"
+    model: "csv"
+    source: "data/FL/regression/vfl/test/host.csv"
   - description: "regression_hfl_train_client1"
     model: "csv"
     source: "data/FL/regression/hfl/train/client1.csv"

config/node2.yaml

@@ -82,6 +82,12 @@ datasets:
     source: "data/FL/multiclass/hfl/test/client2.csv"
 
   # FL regression
+  - description: "regression_vfl_train_guest"
+    model: "csv"
+    source: "data/FL/regression/vfl/train/guest.csv"
+  - description: "regression_vfl_test_guest"
+    model: "csv"
+    source: "data/FL/regression/vfl/test/guest.csv"
   - description: "regression_hfl_train_client2"
     model: "csv"
     source: "data/FL/regression/hfl/train/client2.csv"

config/primihub_node1.yaml

@@ -83,6 +83,12 @@ datasets:
     source: "data/FL/multiclass/hfl/test/client1.csv"
 
   # FL regression
+  - description: "regression_vfl_train_host"
+    model: "csv"
+    source: "data/FL/regression/vfl/train/host.csv"
+  - description: "regression_vfl_test_host"
+    model: "csv"
+    source: "data/FL/regression/vfl/test/host.csv"
   - description: "regression_hfl_train_client1"
     model: "csv"
     source: "data/FL/regression/hfl/train/client1.csv"

config/primihub_node2.yaml

@@ -63,6 +63,12 @@ datasets:
     source: "data/FL/multiclass/hfl/test/client2.csv"
 
   # FL regression
+  - description: "regression_vfl_train_guest"
+    model: "csv"
+    source: "data/FL/regression/vfl/train/guest.csv"
+  - description: "regression_vfl_test_guest"
+    model: "csv"
+    source: "data/FL/regression/vfl/test/guest.csv"
   - description: "regression_hfl_train_client2"
     model: "csv"
     source: "data/FL/regression/hfl/train/client2.csv"

data/FL/regression/vfl/test/guest.csv

+id,CRIM,ZN,INDUS,CHAS,NOX,RM
+79,0.08387,0.0,12.83,0,0.437,5.874
+96,0.11504,0.0,2.89,0,0.445,6.163
+19,0.7258,0.0,8.14,0,0.538,5.727
+371,9.2323,0.0,18.1,0,0.631,6.216
+173,0.09178,0.0,4.05,0,0.51,6.416
+315,0.25356,0.0,9.9,0,0.544,5.705
+16,1.05393,0.0,8.14,0,0.538,5.935
+456,4.66883,0.0,18.1,0,0.713,5.976
+238,0.08244,30.0,4.93,0,0.428,6.481
+219,0.11425,0.0,13.89,1,0.55,6.373
+231,0.46296,0.0,6.2,0,0.504,7.412
+65,0.03584,80.0,3.37,0,0.398,6.29
+88,0.0566,0.0,3.41,0,0.489,7.007
+370,6.53876,0.0,18.1,1,0.631,7.016
+495,0.17899,0.0,9.69,0,0.585,5.67
+145,2.37934,0.0,19.58,0,0.871,6.13
+22,1.23247,0.0,8.14,0,0.538,6.142
+141,1.62864,0.0,21.89,0,0.624,5.019
+286,0.01965,80.0,1.76,0,0.385,6.23
+210,0.17446,0.0,10.59,1,0.489,5.96
+135,0.55778,0.0,21.89,0,0.624,6.335
+42,0.1415,0.0,6.91,0,0.448,6.169
+452,5.09017,0.0,18.1,0,0.713,6.297
+367,13.5222,0.0,18.1,0,0.631,3.863
+199,0.0315,95.0,1.47,0,0.403,6.975
+188,0.12579,45.0,3.44,0,0.437,6.556
+385,16.8118,0.0,18.1,0,0.7,5.277
+243,0.12757,30.0,4.93,0,0.428,6.393
+349,0.02899,40.0,1.25,0,0.429,6.939
+296,0.05372,0.0,13.92,0,0.437,6.549
+406,20.7162,0.0,18.1,0,0.659,4.138
+314,0.3692,0.0,9.9,0,0.544,6.567
+398,38.3518,0.0,18.1,0,0.693,5.453
+221,0.40771,0.0,6.2,1,0.507,6.164
+174,0.08447,0.0,4.05,0,0.51,5.859
+232,0.57529,0.0,6.2,0,0.507,8.337
+92,0.04203,28.0,15.04,0,0.464,6.442
+97,0.12083,0.0,2.89,0,0.445,8.069
+450,6.71772,0.0,18.1,0,0.713,6.749
+108,0.12802,0.0,8.56,0,0.52,6.474
+183,0.10008,0.0,2.46,0,0.488,6.563
+165,2.924,0.0,19.58,0,0.605,6.101
+332,0.03466,35.0,6.06,0,0.4379,6.031
+74,0.07896,0.0,12.83,0,0.437,6.273
+347,0.0187,85.0,4.15,0,0.429,6.516
+204,0.02009,95.0,2.68,0,0.4161,8.034
+192,0.08664,45.0,3.44,0,0.437,7.178
+228,0.29819,0.0,6.2,0,0.504,7.686
+26,0.67191,0.0,8.14,0,0.538,5.813
+310,2.63548,0.0,9.9,0,0.544,4.973
+103,0.21161,0.0,8.56,0,0.52,6.137
+496,0.2896,0.0,9.69,0,0.585,5.39
+215,0.19802,0.0,10.59,0,0.489,6.182
+474,8.05579,0.0,18.1,0,0.584,5.427
+407,11.9511,0.0,18.1,0,0.659,5.608
+80,0.04113,25.0,4.86,0,0.426,6.727
+267,0.57834,20.0,3.97,0,0.575,8.297
+304,0.05515,33.0,2.18,0,0.472,7.236
+431,10.0623,0.0,18.1,0,0.584,6.833
+147,2.36862,0.0,19.58,0,0.871,4.926
+71,0.15876,0.0,10.81,0,0.413,5.961
+212,0.21719,0.0,10.59,1,0.489,5.807
+487,4.83567,0.0,18.1,0,0.583,5.905
+363,4.22239,0.0,18.1,1,0.77,5.803
+223,0.6147,0.0,6.2,0,0.507,6.618
+464,7.83932,0.0,18.1,0,0.655,6.209
+216,0.0456,0.0,13.89,1,0.55,5.888
+291,0.07886,80.0,4.95,0,0.411,7.148
+15,0.62739,0.0,8.14,0,0.538,5.834
+164,2.24236,0.0,19.58,0,0.605,5.854
+244,0.20608,22.0,5.86,0,0.431,5.593
+2,0.02729,0.0,7.07,0,0.469,7.185
+206,0.22969,0.0,10.59,0,0.489,6.326
+112,0.12329,0.0,10.01,0,0.547,5.913
+328,0.06617,0.0,3.24,0,0.46,5.868
+218,0.11069,0.0,13.89,1,0.55,5.951
+397,7.67202,0.0,18.1,0,0.693,5.747
+227,0.41238,0.0,6.2,0,0.504,7.163
+485,3.67367,0.0,18.1,0,0.583,6.312
+46,0.18836,0.0,6.91,0,0.448,5.786
+233,0.33147,0.0,6.2,0,0.507,8.247
+111,0.10084,0.0,10.01,0,0.547,6.715
+68,0.13554,12.5,6.07,0,0.409,5.594
+70,0.08826,0.0,10.81,0,0.413,6.417
+104,0.1396,0.0,8.56,0,0.52,6.167
+479,14.3337,0.0,18.1,0,0.614,6.229
+85,0.05735,0.0,4.49,0,0.449,6.63
+239,0.09252,30.0,4.93,0,0.428,6.606
+144,2.77974,0.0,19.58,0,0.871,4.903
+7,0.14455,12.5,7.87,0,0.524,6.172
+159,1.42502,0.0,19.58,0,0.871,6.51
+253,0.36894,22.0,5.86,0,0.431,8.259
+255,0.03548,80.0,3.64,0,0.392,5.876
+200,0.01778,95.0,1.47,0,0.403,7.135
+116,0.13158,0.0,10.01,0,0.547,6.176
+242,0.1029,30.0,4.93,0,0.428,6.358
+313,0.26938,0.0,9.9,0,0.544,6.266
+120,0.06899,0.0,25.65,0,0.581,5.87
+379,17.8667,0.0,18.1,0,0.671,6.223
+378,23.6482,0.0,18.1,0,0.671,6.38
+462,6.65492,0.0,18.1,0,0.713,6.317
+44,0.12269,0.0,6.91,0,0.448,6.069

data/FL/regression/vfl/test/host.csv

+id,AGE,DIS,RAD,TAX,PTRATIO,B,LSTAT,y
+79,36.6,4.5026,5,398.0,18.7,396.06,9.1,20.3
+96,69.6,3.4952,2,276.0,18.0,391.83,11.34,21.4
+19,69.5,3.7965,4,307.0,21.0,390.95,11.28,18.2
+371,100.0,1.1691,24,666.0,20.2,366.15,9.53,50.0
+173,84.1,2.6463,5,296.0,16.6,395.5,9.04,23.6
+315,77.7,3.945,4,304.0,18.4,396.42,11.5,16.2
+16,29.3,4.4986,4,307.0,21.0,386.85,6.58,23.1
+456,87.9,2.5806,24,666.0,20.2,10.48,19.01,12.7
+238,18.5,6.1899,6,300.0,16.6,379.41,6.36,23.7
+219,92.4,3.3633,5,276.0,16.4,393.74,10.5,23.0
+231,76.9,3.6715,8,307.0,17.4,376.14,5.25,31.7
+65,17.8,6.6115,4,337.0,16.1,396.9,4.67,23.5
+88,86.3,3.4217,2,270.0,17.8,396.9,5.5,23.6
+370,97.5,1.2024,24,666.0,20.2,392.05,2.96,50.0
+495,28.8,2.7986,6,391.0,19.2,393.29,17.6,23.1
+145,100.0,1.4191,5,403.0,14.7,172.91,27.8,13.8
+22,91.7,3.9769,4,307.0,21.0,396.9,18.72,15.2
+141,100.0,1.4394,4,437.0,21.2,396.9,34.41,14.4
+286,31.5,9.0892,1,241.0,18.2,341.6,12.93,20.1
+210,92.1,3.8771,4,277.0,18.6,393.25,17.27,21.7
+135,98.2,2.1107,4,437.0,21.2,394.67,16.96,18.1
+42,6.6,5.7209,3,233.0,17.9,383.37,5.81,25.3
+452,91.8,2.3682,24,666.0,20.2,385.09,17.27,16.1
+367,100.0,1.5106,24,666.0,20.2,131.42,13.33,23.1
+199,15.3,7.6534,3,402.0,17.0,396.9,4.56,34.9
+188,29.1,4.5667,5,398.0,15.2,382.84,4.56,29.8
+385,98.1,1.4261,24,666.0,20.2,396.9,30.81,7.2
+243,7.8,7.0355,6,300.0,16.6,374.71,5.19,23.7
+349,34.5,8.7921,1,335.0,19.7,389.85,5.89,26.6
+296,51.0,5.9604,4,289.0,16.0,392.85,7.39,27.1
+406,100.0,1.1781,24,666.0,20.2,370.22,23.34,11.9
+314,87.3,3.6023,4,304.0,18.4,395.69,9.28,23.8
+398,100.0,1.4896,24,666.0,20.2,396.9,30.59,5.0
+221,91.3,3.048,8,307.0,17.4,395.24,21.46,21.7
+174,68.7,2.7019,5,296.0,16.6,393.23,9.64,22.6
+232,73.3,3.8384,8,307.0,17.4,385.91,2.47,41.7
+92,53.6,3.6659,4,270.0,18.2,395.01,8.16,22.9
+97,76.0,3.4952,2,276.0,18.0,396.9,4.21,38.7
+450,92.6,2.3236,24,666.0,20.2,0.32,17.44,13.4
+108,97.1,2.4329,5,384.0,20.9,395.24,12.27,19.8
+183,95.6,2.847,3,193.0,17.8,396.9,5.68,32.5
+165,93.0,2.2834,5,403.0,14.7,240.16,9.81,25.0
+332,23.3,6.6407,1,304.0,16.9,362.25,7.83,19.4
+74,6.0,4.2515,5,398.0,18.7,394.92,6.78,24.1
+347,27.7,8.5353,4,351.0,17.9,392.43,6.36,23.1
+204,31.9,5.118,4,224.0,14.7,390.55,2.88,50.0
+192,26.3,6.4798,5,398.0,15.2,390.49,2.87,36.4
+228,17.0,3.3751,8,307.0,17.4,377.51,3.92,46.7
+26,90.3,4.682,4,307.0,21.0,376.88,14.81,16.6
+310,37.8,2.5194,4,304.0,18.4,350.45,12.64,16.1
+103,87.4,2.7147,5,384.0,20.9,394.47,13.44,19.3
+496,72.9,2.7986,6,391.0,19.2,396.9,21.14,19.7
+215,42.4,3.9454,4,277.0,18.6,393.63,9.47,25.0
+474,95.4,2.4298,24,666.0,20.2,352.58,18.14,13.8
+407,100.0,1.2852,24,666.0,20.2,332.09,12.13,27.9
+80,33.5,5.4007,4,281.0,19.0,396.9,5.29,28.0
+267,67.0,2.4216,5,264.0,13.0,384.54,7.44,50.0
+304,41.1,4.022,7,222.0,18.4,393.68,6.93,36.1
+431,94.3,2.0882,24,666.0,20.2,81.33,19.69,14.1
+147,95.7,1.4608,5,403.0,14.7,391.71,29.53,14.6
+71,17.5,5.2873,4,305.0,19.2,376.94,9.88,21.7
+212,53.8,3.6526,4,277.0,18.6,390.94,16.03,22.4
+487,53.2,3.1523,24,666.0,20.2,388.22,11.45,20.6
+363,89.0,1.9047,24,666.0,20.2,353.04,14.64,16.8
+223,80.8,3.2721,8,307.0,17.4,396.9,7.6,30.1
+464,65.4,2.9634,24,666.0,20.2,396.9,13.22,21.4
+216,56.0,3.1121,5,276.0,16.4,392.8,13.51,23.3
+291,27.7,5.1167,4,245.0,19.2,396.9,3.56,37.3
+15,56.5,4.4986,4,307.0,21.0,395.62,8.47,19.9
+164,91.8,2.422,5,403.0,14.7,395.11,11.64,22.7
+244,76.5,7.9549,7,330.0,19.1,372.49,12.5,17.6
+2,61.1,4.9671,2,242.0,17.8,392.83,4.03,34.7
+206,52.5,4.3549,4,277.0,18.6,394.87,10.97,24.4
+112,92.9,2.3534,6,432.0,17.8,394.95,16.21,18.8
+328,25.8,5.2146,4,430.0,16.9,382.44,9.97,19.3
+218,93.8,2.8893,5,276.0,16.4,396.9,17.92,21.5
+397,98.9,1.6334,24,666.0,20.2,393.1,19.92,8.5
+227,79.9,3.2157,8,307.0,17.4,372.08,6.36,31.6
+485,51.9,3.9917,24,666.0,20.2,388.62,10.58,21.2
+46,33.3,5.1004,3,233.0,17.9,396.9,14.15,20.0
+233,70.4,3.6519,8,307.0,17.4,378.95,3.95,48.3
+111,81.6,2.6775,6,432.0,17.8,395.59,10.16,22.8
+68,36.8,6.498,4,345.0,18.9,396.9,13.09,17.4
+70,6.6,5.2873,4,305.0,19.2,383.73,6.72,24.2
+104,90.0,2.421,5,384.0,20.9,392.69,12.33,20.1
+479,88.0,1.9512,24,666.0,20.2,383.32,13.11,21.4
+85,56.1,4.4377,3,247.0,18.5,392.3,6.53,26.6
+239,42.2,6.1899,6,300.0,16.6,383.78,7.37,23.3
+144,97.8,1.3459,5,403.0,14.7,396.9,29.29,11.8
+7,96.1,5.9505,5,311.0,15.2,396.9,19.15,27.1
+159,100.0,1.7659,5,403.0,14.7,364.31,7.39,23.3
+253,8.4,8.9067,7,330.0,19.1,396.9,3.54,42.8
+255,19.1,9.2203,1,315.0,16.4,395.18,9.25,20.9
+200,13.9,7.6534,3,402.0,17.0,384.3,4.45,32.9
+116,72.5,2.7301,6,432.0,17.8,393.3,12.04,21.2
+242,52.9,7.0355,6,300.0,16.6,372.75,11.22,22.2
+313,82.8,3.2628,4,304.0,18.4,393.39,7.9,21.6
+120,69.7,2.2577,2,188.0,19.1,389.15,14.37,22.0
+379,100.0,1.3861,24,666.0,20.2,393.74,21.78,10.2
+378,96.2,1.3861,24,666.0,20.2,396.9,23.69,13.1
+462,83.0,2.7344,24,666.0,20.2,396.9,13.99,19.5
+44,40.0,5.7209,3,233.0,17.9,389.39,9.55,21.2

data/FL/regression/vfl/train/guest.csv

+id,CRIM,ZN,INDUS,CHAS,NOX,RM
+91,0.03932,0.0,3.41,0,0.489,6.405
+273,0.22188,20.0,6.96,1,0.464,7.691
+365,4.55587,0.0,18.1,0,0.718,3.561
+270,0.29916,20.0,6.96,0,0.464,5.856
+469,13.0751,0.0,18.1,0,0.58,5.713
+260,0.54011,20.0,3.97,0,0.647,7.203
+248,0.16439,22.0,5.86,0,0.431,6.433
+471,4.03841,0.0,18.1,0,0.532,6.229
+352,0.07244,60.0,1.69,0,0.411,5.884
+211,0.37578,0.0,10.59,1,0.489,5.404
+41,0.12744,0.0,6.91,0,0.448,6.77
+389,8.15174,0.0,18.1,0,0.7,5.39
+466,3.77498,0.0,18.1,0,0.655,5.952
+266,0.7857,20.0,3.97,0,0.647,7.014
+179,0.0578,0.0,2.46,0,0.488,6.98
+36,0.09744,0.0,5.96,0,0.499,5.841
+8,0.21124,12.5,7.87,0,0.524,5.631
+375,19.6091,0.0,18.1,0,0.671,7.313
+257,0.61154,20.0,3.97,0,0.647,8.704
+196,0.04011,80.0,1.52,0,0.404,7.287
+433,5.58107,0.0,18.1,0,0.713,6.436
+10,0.22489,12.5,7.87,0,0.524,6.377
+264,0.55007,20.0,3.97,0,0.647,7.206
+128,0.32543,0.0,21.89,0,0.624,6.431
+102,0.22876,0.0,8.56,0,0.52,6.405
+430,8.49213,0.0,18.1,0,0.584,6.348
+437,15.1772,0.0,18.1,0,0.74,6.152
+424,8.79212,0.0,18.1,0,0.584,5.565
+393,8.64476,0.0,18.1,0,0.693,6.193
+467,4.42228,0.0,18.1,0,0.584,6.003
+345,0.03113,0.0,4.39,0,0.442,6.014
+124,0.09849,0.0,25.65,0,0.581,5.879
+420,11.0874,0.0,18.1,0,0.718,6.411
+34,1.61282,0.0,8.14,0,0.538,6.096
+208,0.13587,0.0,10.59,1,0.489,6.064
+384,20.0849,0.0,18.1,0,0.7,4.368
+374,18.4982,0.0,18.1,0,0.668,4.138
+237,0.51183,0.0,6.2,0,0.507,7.358
+290,0.03502,80.0,4.95,0,0.411,6.861
+25,0.84054,0.0,8.14,0,0.538,5.599
+28,0.77299,0.0,8.14,0,0.538,6.495
+293,0.08265,0.0,13.92,0,0.437,6.127
+395,8.71675,0.0,18.1,0,0.693,6.471
+51,0.04337,21.0,5.64,0,0.439,6.115
+230,0.537,0.0,6.2,0,0.504,5.981
+254,0.04819,80.0,3.64,0,0.392,6.108
+32,1.38799,0.0,8.14,0,0.538,5.95
+93,0.02875,28.0,15.04,0,0.464,6.211
+391,5.29305,0.0,18.1,0,0.7,6.051
+130,0.34006,0.0,21.89,0,0.624,6.458
+150,1.6566,0.0,19.58,0,0.871,6.122
+444,12.8023,0.0,18.1,0,0.74,5.854
+38,0.17505,0.0,5.96,0,0.499,5.966
+425,15.8603,0.0,18.1,0,0.679,5.896
+324,0.34109,0.0,7.38,0,0.493,6.415
+241,0.10612,30.0,4.93,0,0.428,6.095
+388,14.3337,0.0,18.1,0,0.7,4.88
+396,5.87205,0.0,18.1,0,0.693,6.405
+240,0.11329,30.0,4.93,0,0.428,6.897
+190,0.09068,45.0,3.44,0,0.437,6.951
+329,0.06724,0.0,3.24,0,0.46,6.333
+342,0.02498,0.0,1.89,0,0.518,6.54
+256,0.01538,90.0,3.75,0,0.394,7.454
+27,0.95577,0.0,8.14,0,0.538,6.047
+340,0.06151,0.0,5.19,0,0.515,5.968
+473,4.64689,0.0,18.1,0,0.614,6.98
+468,15.5757,0.0,18.1,0,0.58,5.926
+57,0.01432,100.0,1.32,0,0.411,6.816
+139,0.54452,0.0,21.89,0,0.624,6.151
+185,0.06047,0.0,2.46,0,0.488,6.153
+300,0.04417,70.0,2.24,0,0.4,6.871
+335,0.03961,0.0,5.19,0,0.515,6.037
+449,7.52601,0.0,18.1,0,0.713,6.417
+52,0.0536,21.0,5.64,0,0.439,6.511
+489,0.18337,0.0,27.74,0,0.609,5.414
+399,9.91655,0.0,18.1,0,0.693,5.852
+364,3.47428,0.0,18.1,1,0.718,8.78
+0,0.00632,18.0,2.31,0,0.538,6.575
+177,0.05425,0.0,4.05,0,0.51,6.315
+72,0.09164,0.0,10.81,0,0.413,6.065
+125,0.16902,0.0,25.65,0,0.581,5.986
+279,0.21038,20.0,3.33,0,0.4429,6.812
+402,9.59571,0.0,18.1,0,0.693,6.404
+301,0.03537,34.0,6.09,0,0.433,6.59
+213,0.14052,0.0,10.59,0,0.489,6.375
+146,2.15505,0.0,19.58,0,0.871,5.628
+421,7.02259,0.0,18.1,0,0.718,6.006
+372,8.26725,0.0,18.1,1,0.668,5.875
+504,0.10959,0.0,11.93,0,0.573,6.794
+202,0.02177,82.5,2.03,0,0.415,7.61
+438,13.6781,0.0,18.1,0,0.74,5.935
+47,0.22927,0.0,6.91,0,0.448,6.03
+380,88.9762,0.0,18.1,0,0.671,6.968
+343,0.02543,55.0,3.78,0,0.484,6.696
+359,4.26131,0.0,18.1,0,0.77,6.112
+418,73.5341,0.0,18.1,0,0.679,5.957
+40,0.03359,75.0,2.95,0,0.428,7.024
+428,7.36711,0.0,18.1,0,0.679,6.193
+39,0.02763,75.0,2.95,0,0.428,6.595
+224,0.31533,0.0,6.2,0,0.504,8.266
+394,13.3598,0.0,18.1,0,0.693,5.887
+426,12.2472,0.0,18.1,0,0.584,5.837
+502,0.04527,0.0,11.93,0,0.573,6.12
+439,9.39063,0.0,18.1,0,0.74,5.627
+415,18.0846,0.0,18.1,0,0.679,6.434
+37,0.08014,0.0,5.96,0,0.499,5.85
+99,0.0686,0.0,2.89,0,0.445,7.416
+186,0.05602,0.0,2.46,0,0.488,7.831
+234,0.44791,0.0,6.2,1,0.507,6.726
+351,0.0795,60.0,1.69,0,0.411,6.579
+276,0.10469,40.0,6.41,1,0.447,7.267
+476,4.87141,0.0,18.1,0,0.614,6.484
+105,0.13262,0.0,8.56,0,0.52,5.851
+151,1.49632,0.0,19.58,0,0.871,5.404
+262,0.52014,20.0,3.97,0,0.647,8.398
+287,0.03871,52.5,5.32,0,0.405,6.209
+325,0.19186,0.0,7.38,0,0.493,6.431
+414,45.7461,0.0,18.1,0,0.693,4.519
+472,3.56868,0.0,18.1,0,0.58,6.437
+149,2.73397,0.0,19.58,0,0.871,5.597
+412,18.811,0.0,18.1,0,0.597,4.628
+169,2.44953,0.0,19.58,0,0.605,6.402
+123,0.15038,0.0,25.65,0,0.581,5.856
+483,2.81838,0.0,18.1,0,0.532,5.762
+63,0.1265,25.0,5.13,0,0.453,6.762
+417,25.9406,0.0,18.1,0,0.679,5.304
+157,1.22358,0.0,19.58,0,0.605,6.943
+499,0.17783,0.0,9.69,0,0.585,5.569
+225,0.52693,0.0,6.2,0,0.504,8.725
+323,0.28392,0.0,7.38,0,0.493,5.708
+201,0.03445,82.5,2.03,0,0.415,6.162
+446,6.28807,0.0,18.1,0,0.74,6.341
+401,14.2362,0.0,18.1,0,0.693,6.343
+480,5.82401,0.0,18.1,0,0.532,6.242
+259,0.65665,20.0,3.97,0,0.647,6.842
+175,0.06664,0.0,4.05,0,0.51,6.546
+503,0.06076,0.0,11.93,0,0.573,6.976
+245,0.19133,22.0,5.86,0,0.431,5.605
+376,15.288,0.0,18.1,0,0.671,6.649
+55,0.01311,90.0,1.22,0,0.403,7.249
+443,9.96654,0.0,18.1,0,0.74,6.485
+447,9.92485,0.0,18.1,0,0.74,6.251
+119,0.14476,0.0,10.01,0,0.547,5.731
+440,22.0511,0.0,18.1,0,0.74,5.818
+460,4.81213,0.0,18.1,0,0.713,6.701
+333,0.05083,0.0,5.19,0,0.515,6.316
+48,0.25387,0.0,6.91,0,0.448,5.399
+382,9.18702,0.0,18.1,0,0.7,5.536
+9,0.17004,12.5,7.87,0,0.524,6.004
+252,0.08221,22.0,5.86,0,0.431,6.957
+155,3.53501,0.0,19.58,1,0.871,6.152
+366,3.69695,0.0,18.1,0,0.718,4.963
+338,0.03306,0.0,5.19,0,0.515,6.059
+178,0.06642,0.0,4.05,0,0.51,6.86
+493,0.17331,0.0,9.69,0,0.585,5.707
+432,6.44405,0.0,18.1,0,0.584,6.425
+75,0.09512,0.0,12.83,0,0.437,6.286
+229,0.44178,0.0,6.2,0,0.504,6.552
+346,0.06162,0.0,4.39,0,0.442,5.898
+284,0.00906,90.0,2.97,0,0.4,7.088
+194,0.01439,60.0,2.93,0,0.401,6.604
+161,1.46336,0.0,19.58,0,0.605,7.489
+168,2.3004,0.0,19.58,0,0.605,6.319
+269,0.09065,20.0,6.96,1,0.464,5.92
+83,0.03551,25.0,4.86,0,0.426,6.167
+195,0.01381,80.0,0.46,0,0.422,7.875
+129,0.88125,0.0,21.89,0,0.624,5.637
+184,0.08308,0.0,2.46,0,0.488,5.604
+416,10.8342,0.0,18.1,0,0.679,6.782
+498,0.23912,0.0,9.69,0,0.585,6.019
+354,0.04301,80.0,1.91,0,0.413,5.663
+475,6.39312,0.0,18.1,0,0.584,6.162
+403,24.8017,0.0,18.1,0,0.693,5.349
+115,0.17134,0.0,10.01,0,0.547,5.928
+319,0.47547,0.0,9.9,0,0.544,6.113
+302,0.09266,34.0,6.09,0,0.433,6.495
+156,2.44668,0.0,19.58,0,0.871,5.272
+305,0.05479,33.0,2.18,0,0.472,6.616
+486,5.69175,0.0,18.1,0,0.583,6.114
+261,0.53412,20.0,3.97,0,0.647,7.52
+13,0.62976,0.0,8.14,0,0.538,5.949
+427,37.6619,0.0,18.1,0,0.679,6.202
+307,0.04932,33.0,2.18,0,0.472,6.849
+94,0.04294,28.0,15.04,0,0.464,6.249
+69,0.12816,12.5,6.07,0,0.409,5.885
+336,0.03427,0.0,5.19,0,0.515,5.869
+45,0.17142,0.0,6.91,0,0.448,5.682
+334,0.03738,0.0,5.19,0,0.515,6.31
+299,0.05561,70.0,2.24,0,0.4,7.041
+308,0.49298,0.0,9.9,0,0.544,6.635
+131,1.19294,0.0,21.89,0,0.624,6.326
+60,0.14932,25.0,5.13,0,0.453,5.741
+306,0.07503,33.0,2.18,0,0.472,7.42
+457,8.20058,0.0,18.1,0,0.713,5.936
+84,0.05059,0.0,4.49,0,0.449,6.389
+140,0.2909,0.0,21.89,0,0.624,6.174
+58,0.15445,25.0,5.13,0,0.453,6.145
+29,1.00245,0.0,8.14,0,0.538,6.674
+268,0.5405,20.0,3.97,0,0.575,7.47
+220,0.35809,0.0,6.2,1,0.507,6.951
+285,0.01096,55.0,2.25,0,0.389,6.453
+392,11.5779,0.0,18.1,0,0.7,5.036
+500,0.22438,0.0,9.69,0,0.585,6.027
+362,3.67822,0.0,18.1,0,0.77,5.362
+289,0.04297,52.5,5.32,0,0.405,6.565
+377,9.82349,0.0,18.1,0,0.671,6.794
+198,0.03768,80.0,1.52,0,0.404,7.274
+454,9.51363,0.0,18.1,0,0.713,6.728
+89,0.05302,0.0,3.41,0,0.489,7.079
+353,0.01709,90.0,2.02,0,0.41,6.728
+14,0.63796,0.0,8.14,0,0.538,6.096
+95,0.12204,0.0,2.89,0,0.445,6.625
+166,2.01019,0.0,19.58,0,0.605,7.929
+176,0.07022,0.0,4.05,0,0.51,6.02
+337,0.03041,0.0,5.19,0,0.515,5.895
+309,0.3494,0.0,9.9,0,0.544,5.972
+182,0.09103,0.0,2.46,0,0.488,7.155
+294,0.08199,0.0,13.92,0,0.437,6.009
+316,0.31827,0.0,9.9,0,0.544,5.914
+110,0.10793,0.0,8.56,0,0.52,6.195
+247,0.19657,22.0,5.86,0,0.431,6.226
+126,0.38735,0.0,25.65,0,0.581,5.613
+251,0.21409,22.0,5.86,0,0.431,6.438
+497,0.26838,0.0,9.69,0,0.585,5.794
+327,0.24103,0.0,7.38,0,0.493,6.083
+320,0.1676,0.0,7.38,0,0.493,6.426
+163,1.51902,0.0,19.58,1,0.605,8.375
+278,0.07978,40.0,6.41,0,0.447,6.482
+303,0.1,34.0,6.09,0,0.433,6.982
+137,0.35233,0.0,21.89,0,0.624,6.454
+31,1.35472,0.0,8.14,0,0.538,6.072
+505,0.04741,0.0,11.93,0,0.573,6.03
+387,22.5971,0.0,18.1,0,0.7,5.0
+127,0.25915,0.0,21.89,0,0.624,5.693
+136,0.32264,0.0,21.89,0,0.624,5.942
+434,13.9134,0.0,18.1,0,0.713,6.208
+33,1.15172,0.0,8.14,0,0.538,5.701
+203,0.0351,95.0,2.68,0,0.4161,7.853
+53,0.04981,21.0,5.64,0,0.439,5.998
+459,6.80117,0.0,18.1,0,0.713,6.081
+30,1.13081,0.0,8.14,0,0.538,5.713
+62,0.11027,25.0,5.13,0,0.453,6.456
+274,0.05644,40.0,6.41,1,0.447,6.758
+470,4.34879,0.0,18.1,0,0.58,6.167
+298,0.06466,70.0,2.24,0,0.4,6.345
+109,0.26363,0.0,8.56,0,0.52,6.229
+348,0.01501,80.0,2.01,0,0.435,6.635
+50,0.08873,21.0,5.64,0,0.439,5.963
+410,51.1358,0.0,18.1,0,0.597,5.757
+217,0.07013,0.0,13.89,0,0.55,6.642
+64,0.01951,17.5,1.38,0,0.4161,7.104
+501,0.06263,0.0,11.93,0,0.573,6.593
+5,0.02985,0.0,2.18,0,0.458,6.43
+356,8.98296,0.0,18.1,1,0.77,6.212
+121,0.07165,0.0,25.65,0,0.581,6.004
+288,0.0459,52.5,5.32,0,0.405,6.315
+250,0.1403,22.0,5.86,0,0.431,6.487
+189,0.0837,45.0,3.44,0,0.437,7.185
+482,5.73116,0.0,18.1,0,0.532,7.061
+18,0.80271,0.0,8.14,0,0.538,5.456
+180,0.06588,0.0,2.46,0,0.488,7.765
+205,0.13642,0.0,10.59,0,0.489,5.891
+82,0.03659,25.0,4.86,0,0.426,6.302
+404,41.5292,0.0,18.1,0,0.693,5.531
+214,0.28955,0.0,10.59,0,0.489,5.412
+339,0.05497,0.0,5.19,0,0.515,5.985
+81,0.04462,25.0,4.86,0,0.426,6.619
+461,3.69311,0.0,18.1,0,0.713,6.376
+350,0.06211,40.0,1.25,0,0.429,6.49
+458,7.75223,0.0,18.1,0,0.713,6.301
+405,67.9208,0.0,18.1,0,0.693,5.683
+295,0.12932,0.0,13.92,0,0.437,6.678
+341,0.01301,35.0,1.52,0,0.442,7.241
+114,0.14231,0.0,10.01,0,0.547,6.254
+24,0.75026,0.0,8.14,0,0.538,5.924
+481,5.70818,0.0,18.1,0,0.532,6.75
+23,0.98843,0.0,8.14,0,0.538,5.813
+283,0.01501,90.0,1.21,1,0.401,7.923
+77,0.08707,0.0,12.83,0,0.437,6.14
+4,0.06905,0.0,2.18,0,0.458,7.147
+170,1.20742,0.0,19.58,0,0.605,5.875
+318,0.40202,0.0,9.9,0,0.544,6.382
+101,0.11432,0.0,8.56,0,0.52,6.781
+422,12.0482,0.0,18.1,0,0.614,5.648
+258,0.66351,20.0,3.97,0,0.647,7.333
+118,0.13058,0.0,10.01,0,0.547,5.872
+106,0.1712,0.0,8.56,0,0.52,5.836
+107,0.13117,0.0,8.56,0,0.52,6.127
+49,0.21977,0.0,6.91,0,0.448,5.602
+330,0.04544,0.0,3.24,0,0.46,6.144
+281,0.03705,20.0,3.33,0,0.4429,6.968
+98,0.08187,0.0,2.89,0,0.445,7.82
+436,14.4208,0.0,18.1,0,0.74,6.461
+87,0.07151,0.0,4.49,0,0.449,6.121
+368,4.89822,0.0,18.1,0,0.631,4.97
+76,0.10153,0.0,12.83,0,0.437,6.279
+331,0.05023,35.0,6.06,0,0.4379,5.706
+488,0.15086,0.0,27.74,0,0.609,5.454
+153,2.14918,0.0,19.58,0,0.871,5.709
+20,1.25179,0.0,8.14,0,0.538,5.57
+43,0.15936,0.0,6.91,0,0.448,6.211
+154,1.41385,0.0,19.58,1,0.871,6.129
+409,14.4383,0.0,18.1,0,0.597,6.852
+162,1.83377,0.0,19.58,1,0.605,7.802
+265,0.76162,20.0,3.97,0,0.647,5.56
+361,3.83684,0.0,18.1,0,0.77,6.251
+282,0.06129,20.0,3.33,1,0.4429,7.645
+275,0.09604,40.0,6.41,0,0.447,6.854
+59,0.10328,25.0,5.13,0,0.453,5.927
+292,0.03615,80.0,4.95,0,0.411,6.63
+61,0.17171,25.0,5.13,0,0.453,5.966
+56,0.02055,85.0,0.74,0,0.41,6.383
+3,0.03237,0.0,2.18,0,0.458,6.998
+226,0.38214,0.0,6.2,0,0.504,8.04
+197,0.04666,80.0,1.52,0,0.404,7.107
+142,3.32105,0.0,19.58,1,0.871,5.403
+455,4.75237,0.0,18.1,0,0.713,6.525
+172,0.13914,0.0,4.05,0,0.51,5.572
+181,0.06888,0.0,2.46,0,0.488,6.144
+134,0.97617,0.0,21.89,0,0.624,5.757
+477,15.0234,0.0,18.1,0,0.614,5.304
+236,0.52058,0.0,6.2,1,0.507,6.631
+494,0.27957,0.0,9.69,0,0.585,5.926
+280,0.03578,20.0,3.33,0,0.4429,7.82
+152,1.12658,0.0,19.58,1,0.871,5.012
+187,0.07875,45.0,3.44,0,0.437,6.782
+191,0.06911,45.0,3.44,0,0.437,6.739
+209,0.43571,0.0,10.59,1,0.489,5.344
+167,1.80028,0.0,19.58,0,0.605,5.877
+100,0.14866,0.0,8.56,0,0.52,6.727
+445,10.6718,0.0,18.1,0,0.74,6.459
+148,2.33099,0.0,19.58,0,0.871,5.186
+317,0.24522,0.0,9.9,0,0.544,5.782
+390,6.96215,0.0,18.1,0,0.7,5.713
+117,0.15098,0.0,10.01,0,0.547,6.021
+78,0.05646,0.0,12.83,0,0.437,6.232
+369,5.66998,0.0,18.1,1,0.631,6.683
+17,0.7842,0.0,8.14,0,0.538,5.99
+326,0.30347,0.0,7.38,0,0.493,6.312
+441,9.72418,0.0,18.1,0,0.74,6.406
+413,28.6558,0.0,18.1,0,0.597,5.155
+143,4.0974,0.0,19.58,0,0.871,5.468
+272,0.1146,20.0,6.96,0,0.464,6.538
+492,0.11132,0.0,27.74,0,0.609,5.983
+321,0.18159,0.0,7.38,0,0.493,6.376
+171,2.3139,0.0,19.58,0,0.605,5.88
+11,0.11747,12.5,7.87,0,0.524,6.009
+357,3.8497,0.0,18.1,1,0.77,6.395
+322,0.35114,0.0,7.38,0,0.493,6.041
+86,0.05188,0.0,4.49,0,0.449,6.015
+411,14.0507,0.0,18.1,0,0.597,6.657
+160,1.27346,0.0,19.58,1,0.605,6.25
+400,25.0461,0.0,18.1,0,0.693,5.987
+355,0.10659,80.0,1.91,0,0.413,5.936
+297,0.14103,0.0,13.92,0,0.437,5.79
+235,0.33045,0.0,6.2,0,0.507,6.086
+249,0.19073,22.0,5.86,0,0.431,6.718
+429,9.33889,0.0,18.1,0,0.679,6.38
+113,0.22212,0.0,10.01,0,0.547,6.092
+1,0.02731,0.0,7.07,0,0.469,6.421
+358,5.20177,0.0,18.1,1,0.77,6.127
+491,0.10574,0.0,27.74,0,0.609,5.983
+344,0.03049,55.0,3.78,0,0.484,6.874
+193,0.02187,60.0,2.93,0,0.401,6.8
+207,0.25199,0.0,10.59,0,0.489,5.783
+67,0.05789,12.5,6.07,0,0.409,5.878
+246,0.33983,22.0,5.86,0,0.431,6.108
+419,11.8123,0.0,18.1,0,0.718,6.824
+408,7.40389,0.0,18.1,0,0.597,5.617
+132,0.59005,0.0,21.89,0,0.624,6.372
+386,24.3938,0.0,18.1,0,0.7,4.652
+373,11.1081,0.0,18.1,0,0.668,4.906
+484,2.37857,0.0,18.1,0,0.583,5.871
+90,0.04684,0.0,3.41,0,0.489,6.417
+381,15.8744,0.0,18.1,0,0.671,6.545
+122,0.09299,0.0,25.65,0,0.581,5.961
+448,9.32909,0.0,18.1,0,0.713,6.185
+73,0.19539,0.0,10.81,0,0.413,6.245
+6,0.08829,12.5,7.87,0,0.524,6.012
+158,1.34284,0.0,19.58,0,0.605,6.066
+442,5.66637,0.0,18.1,0,0.74,6.219
+383,7.99248,0.0,18.1,0,0.7,5.52
+463,5.82115,0.0,18.1,0,0.713,6.513
+435,11.1604,0.0,18.1,0,0.74,6.629
+465,3.1636,0.0,18.1,0,0.655,5.759
+12,0.09378,12.5,7.87,0,0.524,5.889
+35,0.06417,0.0,5.96,0,0.499,5.933
+478,10.233,0.0,18.1,0,0.614,6.185
+360,4.54192,0.0,18.1,0,0.77,6.398
+277,0.06127,40.0,6.41,1,0.447,6.826
+312,0.26169,0.0,9.9,0,0.544,6.023
+451,5.44114,0.0,18.1,0,0.713,6.655
+133,0.32982,0.0,21.89,0,0.624,5.822
+263,0.82526,20.0,3.97,0,0.647,7.327
+490,0.20746,0.0,27.74,0,0.609,5.093
+66,0.04379,80.0,3.37,0,0.398,5.787
+138,0.2498,0.0,21.89,0,0.624,5.857
+54,0.0136,75.0,4.0,0,0.41,5.888
+271,0.16211,20.0,6.96,0,0.464,6.24
+423,7.05042,0.0,18.1,0,0.614,6.103
+311,0.79041,0.0,9.9,0,0.544,6.122
+21,0.85204,0.0,8.14,0,0.538,5.965
+453,8.24809,0.0,18.1,0,0.713,7.393
+222,0.62356,0.0,6.2,1,0.507,6.879

data/FL/regression/vfl/train/host.csv

+id,AGE,DIS,RAD,TAX,PTRATIO,B,LSTAT,y
+91,73.9,3.0921,2,270.0,17.8,393.55,8.2,22.0
+273,51.8,4.3665,3,223.0,18.6,390.77,6.58,35.2
+365,87.9,1.6132,24,666.0,20.2,354.7,7.12,27.5
+270,42.1,4.429,3,223.0,18.6,388.65,13.0,21.1
+469,56.7,2.8237,24,666.0,20.2,396.9,14.76,20.1
+260,81.8,2.1121,5,264.0,13.0,392.8,9.59,33.8
+248,49.1,7.8265,7,330.0,19.1,374.71,9.52,24.5
+471,90.7,3.0993,24,666.0,20.2,395.33,12.87,19.6
+352,18.5,10.7103,4,411.0,18.3,392.33,7.79,18.6
+211,88.6,3.665,4,277.0,18.6,395.24,23.98,19.3
+41,2.9,5.7209,3,233.0,17.9,385.41,4.84,26.6
+389,98.9,1.7281,24,666.0,20.2,396.9,20.85,11.5
+466,84.7,2.8715,24,666.0,20.2,22.01,17.15,19.0
+266,84.6,2.1329,5,264.0,13.0,384.07,14.79,30.7
+179,58.4,2.829,3,193.0,17.8,396.9,5.04,37.2
+36,61.4,3.3779,5,279.0,19.2,377.56,11.41,20.0
+8,100.0,6.0821,5,311.0,15.2,386.63,29.93,16.5
+375,97.9,1.3163,24,666.0,20.2,396.9,13.44,15.0
+257,86.9,1.801,5,264.0,13.0,389.7,5.12,50.0
+196,34.1,7.309,2,329.0,12.6,396.9,4.08,33.3
+433,87.9,2.3158,24,666.0,20.2,100.19,16.22,14.3
+10,94.3,6.3467,5,311.0,15.2,392.52,20.45,15.0
+264,91.6,1.9301,5,264.0,13.0,387.89,8.1,36.5
+128,98.8,1.8125,4,437.0,21.2,396.9,15.39,18.0
+102,85.4,2.7147,5,384.0,20.9,70.8,10.63,18.6
+430,86.1,2.0527,24,666.0,20.2,83.45,17.64,14.5
+437,100.0,1.9142,24,666.0,20.2,9.32,26.45,8.7
+424,70.6,2.0635,24,666.0,20.2,3.65,17.16,11.7
+393,92.6,1.7912,24,666.0,20.2,396.9,15.17,13.8
+467,94.5,2.5403,24,666.0,20.2,331.29,21.32,19.1
+345,48.5,8.0136,3,352.0,18.8,385.64,10.53,17.5
+124,95.8,2.0063,2,188.0,19.1,379.38,17.58,18.8
+420,100.0,1.8589,24,666.0,20.2,318.75,15.02,16.7
+34,96.9,3.7598,4,307.0,21.0,248.31,20.34,13.5
+208,59.1,4.2392,4,277.0,18.6,381.32,14.66,24.4
+384,91.2,1.4395,24,666.0,20.2,285.83,30.63,8.8
+374,100.0,1.137,24,666.0,20.2,396.9,37.97,13.8
+237,71.6,4.148,8,307.0,17.4,390.07,4.73,31.5
+290,27.9,5.1167,4,245.0,19.2,396.9,3.33,28.5
+25,85.7,4.4546,4,307.0,21.0,303.42,16.51,13.9
+28,94.4,4.4547,4,307.0,21.0,387.94,12.8,18.4
+293,18.4,5.5027,4,289.0,16.0,396.9,8.58,23.9
+395,98.8,1.7257,24,666.0,20.2,391.98,17.12,13.1
+51,63.0,6.8147,4,243.0,16.8,393.97,9.43,20.5
+230,68.1,3.6715,8,307.0,17.4,378.35,11.65,24.3
+254,32.0,9.2203,1,315.0,16.4,392.89,6.57,21.9
+32,82.0,3.99,4,307.0,21.0,232.6,27.71,13.2
+93,28.9,3.6659,4,270.0,18.2,396.33,6.21,25.0
+391,82.5,2.1678,24,666.0,20.2,378.38,18.76,23.2
+130,98.9,2.1185,4,437.0,21.2,395.04,12.6,19.2
+150,97.3,1.618,5,403.0,14.7,372.8,14.1,21.5
+444,96.6,1.8956,24,666.0,20.2,240.52,23.79,10.8
+38,30.2,3.8473,5,279.0,19.2,393.43,10.13,24.7
+425,95.4,1.9096,24,666.0,20.2,7.68,24.39,8.3
+324,40.1,4.7211,5,287.0,19.6,396.9,6.12,25.0
+241,65.1,6.3361,6,300.0,16.6,394.62,12.4,20.1
+388,100.0,1.5895,24,666.0,20.2,372.92,30.62,10.2
+396,96.0,1.6768,24,666.0,20.2,396.9,19.37,12.5
+240,54.3,6.3361,6,300.0,16.6,391.25,11.38,22.0
+190,21.5,6.4798,5,398.0,15.2,377.68,5.1,37.0
+329,17.2,5.2146,4,430.0,16.9,375.21,7.34,22.6
+342,59.7,6.2669,1,422.0,15.9,389.96,8.65,16.5
+256,34.2,6.3361,3,244.0,15.9,386.34,3.11,44.0
+27,88.8,4.4534,4,307.0,21.0,306.38,17.28,14.8
+340,58.5,4.8122,5,224.0,20.2,396.9,9.29,18.7
+473,67.6,2.5329,24,666.0,20.2,374.68,11.66,29.8
+468,71.0,2.9084,24,666.0,20.2,368.74,18.13,19.1
+57,40.5,8.3248,5,256.0,15.1,392.9,3.95,31.6
+139,97.9,1.6687,4,437.0,21.2,396.9,18.46,17.8
+185,68.8,3.2797,3,193.0,17.8,387.11,13.15,29.6
+300,47.4,7.8278,5,358.0,14.8,390.86,6.07,24.8
+335,34.5,5.9853,5,224.0,20.2,396.9,8.01,21.1
+449,98.3,2.185,24,666.0,20.2,304.21,19.31,13.0
+52,21.1,6.8147,4,243.0,16.8,396.9,5.28,25.0
+489,98.3,1.7554,4,711.0,20.1,344.05,23.97,7.0
+399,77.8,1.5004,24,666.0,20.2,338.16,29.97,6.3
+364,82.9,1.9047,24,666.0,20.2,354.55,5.29,21.9
+0,65.2,4.09,1,296.0,15.3,396.9,4.98,24.0
+177,73.4,3.3175,5,296.0,16.6,395.6,6.29,24.6
+72,7.8,5.2873,4,305.0,19.2,390.91,5.52,22.8
+125,88.4,1.9929,2,188.0,19.1,385.02,14.81,21.4
+279,32.2,4.1007,5,216.0,14.9,396.9,4.85,35.1
+402,100.0,1.639,24,666.0,20.2,376.11,20.31,12.1
+301,40.4,5.4917,7,329.0,16.1,395.75,9.5,22.0
+213,32.3,3.9454,4,277.0,18.6,385.81,9.38,28.1
+146,100.0,1.5166,5,403.0,14.7,169.27,16.65,15.6
+421,95.3,1.8746,24,666.0,20.2,319.98,15.7,14.2
+372,89.6,1.1296,24,666.0,20.2,347.88,8.88,50.0
+504,89.3,2.3889,1,273.0,21.0,393.45,6.48,22.0
+202,15.7,6.27,2,348.0,14.7,395.38,3.11,42.3
+438,87.9,1.8206,24,666.0,20.2,68.95,34.02,8.4
+47,85.5,5.6894,3,233.0,17.9,392.74,18.8,16.6
+380,91.9,1.4165,24,666.0,20.2,396.9,17.21,10.4
+343,56.4,5.7321,5,370.0,17.6,396.9,7.18,23.9
+359,81.3,2.5091,24,666.0,20.2,390.74,12.67,22.6
+418,100.0,1.8026,24,666.0,20.2,16.45,20.62,8.8
+40,15.8,5.4011,3,252.0,18.3,395.62,1.98,34.9
+428,78.1,1.9356,24,666.0,20.2,96.73,21.52,11.0
+39,21.8,5.4011,3,252.0,18.3,395.63,4.32,30.8
+224,78.3,2.8944,8,307.0,17.4,385.05,4.14,44.8
+394,94.7,1.7821,24,666.0,20.2,396.9,16.35,12.7
+426,59.7,1.9976,24,666.0,20.2,24.65,15.69,10.2
+502,76.7,2.2875,1,273.0,21.0,396.9,9.08,20.6
+439,93.9,1.8172,24,666.0,20.2,396.9,22.88,12.8
+415,100.0,1.8347,24,666.0,20.2,27.25,29.05,7.2
+37,41.5,3.9342,5,279.0,19.2,396.9,8.77,21.0
+99,62.5,3.4952,2,276.0,18.0,396.9,6.19,33.2
+186,53.6,3.1992,3,193.0,17.8,392.63,4.45,50.0
+234,66.5,3.6519,8,307.0,17.4,360.2,8.05,29.0
+351,35.9,10.7103,4,411.0,18.3,370.78,5.49,24.1
+276,49.0,4.7872,4,254.0,17.6,389.25,6.05,33.2
+476,93.6,2.3053,24,666.0,20.2,396.21,18.68,16.7
+105,96.7,2.1069,5,384.0,20.9,394.05,16.47,19.5
+151,100.0,1.5916,5,403.0,14.7,341.6,13.28,19.6
+262,91.5,2.2885,5,264.0,13.0,386.86,5.91,48.8
+287,31.3,7.3172,6,293.0,16.6,396.9,7.14,23.2
+325,14.7,5.4159,5,287.0,19.6,393.68,5.08,24.6
+414,100.0,1.6582,24,666.0,20.2,88.27,36.98,7.0
+472,75.0,2.8965,24,666.0,20.2,393.37,14.36,23.2
+149,94.9,1.5257,5,403.0,14.7,351.85,21.45,15.4
+412,100.0,1.5539,24,666.0,20.2,28.79,34.37,17.9
+169,95.2,2.2625,5,403.0,14.7,330.04,11.32,22.3
+123,97.0,1.9444,2,188.0,19.1,370.31,25.41,17.3
+483,40.3,4.0983,24,666.0,20.2,392.92,10.42,21.8
+63,43.4,7.9809,8,284.0,19.7,395.58,9.5,25.0
+417,89.1,1.6475,24,666.0,20.2,127.36,26.64,10.4
+157,97.4,1.8773,5,403.0,14.7,363.43,4.59,41.3
+499,73.5,2.3999,6,391.0,19.2,395.77,15.1,17.5
+225,83.0,2.8944,8,307.0,17.4,382.0,4.63,50.0
+323,74.3,4.7211,5,287.0,19.6,391.13,11.74,18.5
+201,38.4,6.27,2,348.0,14.7,393.77,7.43,24.1
+446,96.4,2.072,24,666.0,20.2,318.01,17.79,14.9
+401,100.0,1.5741,24,666.0,20.2,396.9,20.32,7.2
+480,64.7,3.4242,24,666.0,20.2,396.9,10.74,23.0
+259,100.0,2.0107,5,264.0,13.0,391.93,6.9,30.1
+175,33.1,3.1323,5,296.0,16.6,390.96,5.33,29.4
+503,91.0,2.1675,1,273.0,21.0,396.9,5.64,23.9
+245,70.2,7.9549,7,330.0,19.1,389.13,18.46,18.5
+376,93.3,1.3449,24,666.0,20.2,363.02,23.24,13.9
+55,21.9,8.6966,5,226.0,17.9,395.93,4.81,35.4
+443,100.0,1.9784,24,666.0,20.2,386.73,18.85,15.4
+447,96.6,2.198,24,666.0,20.2,388.52,16.44,12.6
+119,65.2,2.7592,6,432.0,17.8,391.5,13.61,19.3
+440,92.4,1.8662,24,666.0,20.2,391.45,22.11,10.5
+460,90.0,2.5975,24,666.0,20.2,255.23,16.42,16.4
+333,38.1,6.4584,5,224.0,20.2,389.71,5.68,22.2
+48,95.3,5.87,3,233.0,17.9,396.9,30.81,14.4
+382,100.0,1.5804,24,666.0,20.2,396.9,23.6,11.3
+9,85.9,6.5921,5,311.0,15.2,386.71,17.1,18.9
+252,6.8,8.9067,7,330.0,19.1,386.09,3.53,29.6
+155,82.6,1.7455,5,403.0,14.7,88.01,15.02,15.6
+366,91.4,1.7523,24,666.0,20.2,316.03,14.0,21.9
+338,37.3,4.8122,5,224.0,20.2,396.14,8.51,20.6
+178,74.4,2.9153,5,296.0,16.6,391.27,6.92,29.9
+493,54.0,2.3817,6,391.0,19.2,396.9,12.01,21.8
+432,74.8,2.2004,24,666.0,20.2,97.95,12.03,16.1
+75,45.0,4.5026,5,398.0,18.7,383.23,8.94,21.4
+229,21.4,3.3751,8,307.0,17.4,380.34,3.76,31.5
+346,52.3,8.0136,3,352.0,18.8,364.61,12.67,17.2
+284,20.8,7.3073,1,285.0,15.3,394.72,7.85,32.2
+194,18.8,6.2196,1,265.0,15.6,376.7,4.38,29.1
+161,90.8,1.9709,5,403.0,14.7,374.43,1.73,50.0
+168,96.1,2.1,5,403.0,14.7,297.09,11.1,23.8
+269,61.5,3.9175,3,223.0,18.6,391.34,13.65,20.7
+83,46.7,5.4007,4,281.0,19.0,390.64,7.51,22.9
+195,32.0,5.6484,4,255.0,14.4,394.23,2.97,50.0
+129,94.7,1.9799,4,437.0,21.2,396.9,18.34,14.3
+184,89.8,2.9879,3,193.0,17.8,391.0,13.98,26.4
+416,90.8,1.8195,24,666.0,20.2,21.57,25.79,7.5
+498,65.3,2.4091,6,391.0,19.2,396.9,12.92,21.2
+354,21.9,10.5857,4,334.0,22.0,382.8,8.05,18.2
+475,97.4,2.206,24,666.0,20.2,302.76,24.1,13.3
+403,96.0,1.7028,24,666.0,20.2,396.9,19.77,8.3
+115,88.2,2.4631,6,432.0,17.8,344.91,15.76,18.3
+319,58.8,4.0019,4,304.0,18.4,396.23,12.73,21.0
+302,18.4,5.4917,7,329.0,16.1,383.61,8.67,26.4
+156,94.0,1.7364,5,403.0,14.7,88.63,16.14,13.1
+305,58.1,3.37,7,222.0,18.4,393.36,8.93,28.4
+486,79.8,3.5459,24,666.0,20.2,392.68,14.98,19.1
+261,89.4,2.1398,5,264.0,13.0,388.37,7.26,43.1
+13,61.8,4.7075,4,307.0,21.0,396.9,8.26,20.4
+427,78.7,1.8629,24,666.0,20.2,18.82,14.52,10.9
+307,70.3,3.1827,7,222.0,18.4,396.9,7.53,28.2
+94,77.3,3.615,4,270.0,18.2,396.9,10.59,20.6
+69,33.0,6.498,4,345.0,18.9,396.9,8.79,20.9
+336,46.3,5.2311,5,224.0,20.2,396.9,9.8,19.5
+45,33.8,5.1004,3,233.0,17.9,396.9,10.21,19.3
+334,38.5,6.4584,5,224.0,20.2,389.4,6.75,20.7
+299,10.0,7.8278,5,358.0,14.8,371.58,4.74,29.0
+308,82.5,3.3175,4,304.0,18.4,396.9,4.54,22.8
+131,97.7,2.271,4,437.0,21.2,396.9,12.26,19.6
+60,66.2,7.2254,8,284.0,19.7,395.11,13.15,18.7
+306,71.9,3.0992,7,222.0,18.4,396.9,6.47,33.4
+457,80.3,2.7792,24,666.0,20.2,3.5,16.94,13.5
+84,48.0,4.7794,3,247.0,18.5,396.9,9.62,23.9
+140,93.6,1.6119,4,437.0,21.2,388.08,24.16,14.0
+58,29.2,7.8148,8,284.0,19.7,390.68,6.86,23.3
+29,87.3,4.239,4,307.0,21.0,380.23,11.98,21.0
+268,52.6,2.872,5,264.0,13.0,390.3,3.16,43.5
+220,88.5,2.8617,8,307.0,17.4,391.7,9.71,26.7
+285,31.9,7.3073,1,300.0,15.3,394.72,8.23,22.0
+392,97.0,1.77,24,666.0,20.2,396.9,25.68,9.7
+500,79.7,2.4982,6,391.0,19.2,396.9,14.33,16.8
+362,96.2,2.1036,24,666.0,20.2,380.79,10.19,20.8
+289,22.9,7.3172,6,293.0,16.6,371.72,9.51,24.8
+377,98.8,1.358,24,666.0,20.2,396.9,21.24,13.3
+198,38.3,7.309,2,329.0,12.6,392.2,6.62,34.6
+454,94.1,2.4961,24,666.0,20.2,6.68,18.71,14.9
+89,63.1,3.4145,2,270.0,17.8,396.06,5.7,28.7
+353,36.1,12.1265,5,187.0,17.0,384.46,4.5,30.1
+14,84.5,4.4619,4,307.0,21.0,380.02,10.26,18.2
+95,57.8,3.4952,2,276.0,18.0,357.98,6.65,28.4
+166,96.2,2.0459,5,403.0,14.7,369.3,3.7,50.0
+176,47.2,3.5549,5,296.0,16.6,393.23,10.11,23.2
+337,59.6,5.615,5,224.0,20.2,394.81,10.56,18.5
+309,76.7,3.1025,4,304.0,18.4,396.24,9.97,20.3
+182,92.2,2.7006,3,193.0,17.8,394.12,4.82,37.9
+294,42.3,5.5027,4,289.0,16.0,396.9,10.4,21.7
+316,83.2,3.9986,4,304.0,18.4,390.7,18.33,17.8
+110,54.4,2.7778,5,384.0,20.9,393.49,13.0,21.7
+247,79.2,8.0555,7,330.0,19.1,376.14,10.15,20.5
+126,95.6,1.7572,2,188.0,19.1,359.29,27.26,15.7
+251,8.9,7.3967,7,330.0,19.1,377.07,3.59,24.8
+497,70.6,2.8927,6,391.0,19.2,396.9,14.1,18.3
+327,43.7,5.4159,5,287.0,19.6,396.9,12.79,22.2
+320,52.3,4.5404,5,287.0,19.6,396.9,7.2,23.8
+163,93.9,2.162,5,403.0,14.7,388.45,3.32,50.0
+278,32.1,4.1403,4,254.0,17.6,396.9,7.19,29.1
+303,17.7,5.4917,7,329.0,16.1,390.43,4.86,33.1
+137,98.4,1.8498,4,437.0,21.2,394.08,14.59,17.1
+31,100.0,4.175,4,307.0,21.0,376.73,13.04,14.5
+505,80.8,2.505,1,273.0,21.0,396.9,7.88,11.9
+387,89.5,1.5184,24,666.0,20.2,396.9,31.99,7.4
+127,96.0,1.7883,4,437.0,21.2,392.11,17.19,16.2
+136,93.5,1.9669,4,437.0,21.2,378.25,16.9,17.4
+434,95.0,2.2222,24,666.0,20.2,100.63,15.17,11.7
+33,95.0,3.7872,4,307.0,21.0,358.77,18.35,13.1
+203,33.2,5.118,4,224.0,14.7,392.78,3.81,48.5
+53,21.4,6.8147,4,243.0,16.8,396.9,8.43,23.4
+459,84.4,2.7175,24,666.0,20.2,396.9,14.7,20.0
+30,94.1,4.233,4,307.0,21.0,360.17,22.6,12.7
+62,67.8,7.2255,8,284.0,19.7,396.9,6.73,22.2
+274,32.9,4.0776,4,254.0,17.6,396.9,3.53,32.4
+470,84.0,3.0334,24,666.0,20.2,396.9,16.29,19.9
+298,20.1,7.8278,5,358.0,14.8,368.24,4.97,22.5
+109,91.2,2.5451,5,384.0,20.9,391.23,15.55,19.4
+348,29.7,8.344,4,280.0,17.0,390.94,5.99,24.5
+50,45.7,6.8147,4,243.0,16.8,395.56,13.45,19.7
+410,100.0,1.413,24,666.0,20.2,2.6,10.11,15.0
+217,85.1,3.4211,5,276.0,16.4,392.78,9.69,28.7
+64,59.5,9.2229,3,216.0,18.6,393.24,8.05,33.0
+501,69.1,2.4786,1,273.0,21.0,391.99,9.67,22.4
+5,58.7,6.0622,3,222.0,18.7,394.12,5.21,28.7
+356,97.4,2.1222,24,666.0,20.2,377.73,17.6,17.8
+121,84.1,2.1974,2,188.0,19.1,377.67,14.27,20.3
+288,45.6,7.3172,6,293.0,16.6,396.9,7.6,22.3
+250,13.0,7.3967,7,330.0,19.1,396.28,5.9,24.4
+189,38.9,4.5667,5,398.0,15.2,396.9,5.39,34.9
+482,77.0,3.4106,24,666.0,20.2,395.28,7.01,25.0
+18,36.6,3.7965,4,307.0,21.0,288.99,11.69,20.2
+180,83.3,2.741,3,193.0,17.8,395.56,7.56,39.8
+205,22.3,3.9454,4,277.0,18.6,396.9,10.87,22.6
+82,32.2,5.4007,4,281.0,19.0,396.9,6.72,24.8
+404,85.4,1.6074,24,666.0,20.2,329.46,27.38,8.5
+214,9.8,3.5875,4,277.0,18.6,348.93,29.55,23.7
+339,45.4,4.8122,5,224.0,20.2,396.9,9.74,19.0
+81,70.4,5.4007,4,281.0,19.0,395.63,7.22,23.9
+461,88.4,2.5671,24,666.0,20.2,391.43,14.65,17.7
+350,44.4,8.7921,1,335.0,19.7,396.9,5.98,22.9
+458,83.7,2.7831,24,666.0,20.2,272.21,16.23,14.9
+405,100.0,1.4254,24,666.0,20.2,384.97,22.98,5.0
+295,31.1,5.9604,4,289.0,16.0,396.9,6.27,28.6
+341,49.3,7.0379,1,284.0,15.5,394.74,5.49,32.7
+114,84.2,2.2565,6,432.0,17.8,388.74,10.45,18.5
+24,94.1,4.3996,4,307.0,21.0,394.33,16.3,15.6
+481,74.9,3.3317,24,666.0,20.2,393.07,7.74,23.7
+23,100.0,4.0952,4,307.0,21.0,394.54,19.88,14.5
+283,24.8,5.885,1,198.0,13.6,395.52,3.16,50.0
+77,45.8,4.0905,5,398.0,18.7,386.96,10.27,20.8
+4,54.2,6.0622,3,222.0,18.7,396.9,5.33,36.2
+170,94.6,2.4259,5,403.0,14.7,292.29,14.43,17.4
+318,67.2,3.5325,4,304.0,18.4,395.21,10.36,23.1
+101,71.3,2.8561,5,384.0,20.9,395.58,7.67,26.5
+422,87.6,1.9512,24,666.0,20.2,291.55,14.1,20.8
+258,100.0,1.8946,5,264.0,13.0,383.29,7.79,36.0
+118,73.1,2.4775,6,432.0,17.8,338.63,15.37,20.4
+106,91.9,2.211,5,384.0,20.9,395.67,18.66,19.5
+107,85.2,2.1224,5,384.0,20.9,387.69,14.09,20.4
+49,62.0,6.0877,3,233.0,17.9,396.9,16.2,19.4
+330,32.2,5.8736,4,430.0,16.9,368.57,9.09,19.8
+281,37.2,5.2447,5,216.0,14.9,392.23,4.59,35.4
+98,36.9,3.4952,2,276.0,18.0,393.53,3.57,43.8
+436,93.3,2.0026,24,666.0,20.2,27.49,18.05,9.6
+87,56.8,3.7476,3,247.0,18.5,395.15,8.44,22.2
+368,100.0,1.3325,24,666.0,20.2,375.52,3.26,50.0
+76,74.5,4.0522,5,398.0,18.7,373.66,11.97,20.0
+331,28.4,6.6407,1,304.0,16.9,394.02,12.43,17.1
+488,92.7,1.8209,4,711.0,20.1,395.09,18.06,15.2
+153,98.5,1.6232,5,403.0,14.7,261.95,15.79,19.4
+20,98.1,3.7979,4,307.0,21.0,376.57,21.02,13.6
+43,6.5,5.7209,3,233.0,17.9,394.46,7.44,24.7
+154,96.0,1.7494,5,403.0,14.7,321.02,15.12,17.0
+409,100.0,1.4655,24,666.0,20.2,179.36,19.78,27.5
+162,98.2,2.0407,5,403.0,14.7,389.61,1.92,50.0
+265,62.8,1.9865,5,264.0,13.0,392.4,10.45,22.8
+361,91.1,2.2955,24,666.0,20.2,350.65,14.19,19.9
+282,49.7,5.2119,5,216.0,14.9,377.07,3.01,46.0
+275,42.8,4.2673,4,254.0,17.6,396.9,2.98,32.0
+59,47.2,6.932,8,284.0,19.7,396.9,9.22,19.6
+292,23.4,5.1167,4,245.0,19.2,396.9,4.7,27.9
+61,93.4,6.8185,8,284.0,19.7,378.08,14.44,16.0
+56,35.7,9.1876,2,313.0,17.3,396.9,5.77,24.7
+3,45.8,6.0622,3,222.0,18.7,394.63,2.94,33.4
+226,86.5,3.2157,8,307.0,17.4,387.38,3.13,37.6
+197,36.6,7.309,2,329.0,12.6,354.31,8.61,30.3
+142,100.0,1.3216,5,403.0,14.7,396.9,26.82,13.4
+455,86.5,2.4358,24,666.0,20.2,50.92,18.13,14.1
+172,88.5,2.5961,5,296.0,16.6,396.9,14.69,23.1
+181,62.2,2.5979,3,193.0,17.8,396.9,9.45,36.2
+134,98.4,2.346,4,437.0,21.2,262.76,17.31,15.6
+477,97.3,2.1007,24,666.0,20.2,349.48,24.91,12.0
+236,76.5,4.148,8,307.0,17.4,388.45,9.54,25.1
+494,42.6,2.3817,6,391.0,19.2,396.9,13.59,24.5
+280,64.5,4.6947,5,216.0,14.9,387.31,3.76,45.4
+152,88.0,1.6102,5,403.0,14.7,343.28,12.12,15.3
+187,41.1,3.7886,5,398.0,15.2,393.87,6.68,32.0
+191,30.8,6.4798,5,398.0,15.2,389.71,4.69,30.5
+209,100.0,3.875,4,277.0,18.6,396.9,23.09,20.0
+167,79.2,2.4259,5,403.0,14.7,227.61,12.14,23.8
+100,79.9,2.7778,5,384.0,20.9,394.76,9.42,27.5
+445,94.8,1.9879,24,666.0,20.2,43.06,23.98,11.8
+148,93.8,1.5296,5,403.0,14.7,356.99,28.32,17.8
+317,71.7,4.0317,4,304.0,18.4,396.9,15.94,19.8
+390,97.0,1.9265,24,666.0,20.2,394.43,17.11,15.1
+117,82.6,2.7474,6,432.0,17.8,394.51,10.3,19.2
+78,53.7,5.0141,5,398.0,18.7,386.4,12.34,21.2
+369,96.8,1.3567,24,666.0,20.2,375.33,3.73,50.0
+17,81.7,4.2579,4,307.0,21.0,386.75,14.67,17.5
+326,28.9,5.4159,5,287.0,19.6,396.9,6.15,23.0
+441,97.2,2.0651,24,666.0,20.2,385.96,19.52,17.1
+413,100.0,1.5894,24,666.0,20.2,210.97,20.08,16.3
+143,100.0,1.4118,5,403.0,14.7,396.9,26.42,15.6
+272,58.7,3.9175,3,223.0,18.6,394.96,7.73,24.4
+492,83.5,2.1099,4,711.0,20.1,396.9,13.35,20.1
+321,54.3,4.5404,5,287.0,19.6,396.9,6.87,23.1
+171,97.3,2.3887,5,403.0,14.7,348.13,12.03,19.1
+11,82.9,6.2267,5,311.0,15.2,396.9,13.27,18.9
+357,91.0,2.5052,24,666.0,20.2,391.34,13.27,21.7
+322,49.9,4.7211,5,287.0,19.6,396.9,7.7,20.4
+86,45.1,4.4272,3,247.0,18.5,395.99,12.86,22.5
+411,100.0,1.5275,24,666.0,20.2,35.05,21.22,17.2
+160,92.6,1.7984,5,403.0,14.7,338.92,5.5,27.0
+400,100.0,1.5888,24,666.0,20.2,396.9,26.77,5.6
+355,19.5,10.5857,4,334.0,22.0,376.04,5.57,20.6
+297,58.0,6.32,4,289.0,16.0,396.9,15.84,20.3
+235,61.5,3.6519,8,307.0,17.4,376.75,10.88,24.0
+249,17.5,7.8265,7,330.0,19.1,393.74,6.56,26.2
+429,95.6,1.9682,24,666.0,20.2,60.72,24.08,9.5
+113,95.4,2.548,6,432.0,17.8,396.9,17.09,18.7
+1,78.9,4.9671,2,242.0,17.8,396.9,9.14,21.6
+358,83.4,2.7227,24,666.0,20.2,395.43,11.48,22.7
+491,98.8,1.8681,4,711.0,20.1,390.11,18.07,13.6
+344,28.1,6.4654,5,370.0,17.6,387.97,4.61,31.2
+193,9.9,6.2196,1,265.0,15.6,393.37,5.03,31.1
+207,72.7,4.3549,4,277.0,18.6,389.43,18.06,22.5
+67,21.4,6.498,4,345.0,18.9,396.21,8.1,22.0
+246,34.9,8.0555,7,330.0,19.1,390.18,9.16,24.3
+419,76.5,1.794,24,666.0,20.2,48.45,22.74,8.4
+408,97.9,1.4547,24,666.0,20.2,314.64,26.4,17.2
+132,97.9,2.3274,4,437.0,21.2,385.76,11.12,23.0
+386,100.0,1.4672,24,666.0,20.2,396.9,28.28,10.5
+373,100.0,1.1742,24,666.0,20.2,396.9,34.77,13.8
+484,41.9,3.724,24,666.0,20.2,370.73,13.34,20.6
+90,66.1,3.0923,2,270.0,17.8,392.18,8.81,22.6
+381,99.1,1.5192,24,666.0,20.2,396.9,21.08,10.9
+122,92.9,2.0869,2,188.0,19.1,378.09,17.93,20.5
+448,98.7,2.2616,24,666.0,20.2,396.9,18.13,14.1
+73,6.2,5.2873,4,305.0,19.2,377.17,7.54,23.4
+6,66.6,5.5605,5,311.0,15.2,395.6,12.43,22.9
+158,100.0,1.7573,5,403.0,14.7,353.89,6.43,24.3
+442,100.0,2.0048,24,666.0,20.2,395.69,16.59,18.4
+383,100.0,1.5331,24,666.0,20.2,396.9,24.56,12.3
+463,89.9,2.8016,24,666.0,20.2,393.82,10.29,20.2
+435,94.6,2.1247,24,666.0,20.2,109.85,23.27,13.4
+465,48.2,3.0665,24,666.0,20.2,334.4,14.13,19.9
+12,39.0,5.4509,5,311.0,15.2,390.5,15.71,21.7
+35,68.2,3.3603,5,279.0,19.2,396.9,9.68,18.9
+478,96.7,2.1705,24,666.0,20.2,379.7,18.03,14.6
+360,88.0,2.5182,24,666.0,20.2,374.56,7.79,25.0
+277,27.6,4.8628,4,254.0,17.6,393.45,4.16,33.1
+312,90.4,2.834,4,304.0,18.4,396.3,11.72,19.4
+451,98.2,2.3552,24,666.0,20.2,355.29,17.73,15.2
+133,95.4,2.4699,4,437.0,21.2,388.69,15.03,18.4
+263,94.5,2.0788,5,264.0,13.0,393.42,11.25,31.0
+490,98.0,1.8226,4,711.0,20.1,318.43,29.68,8.1
+66,31.1,6.6115,4,337.0,16.1,396.9,10.24,19.4
+138,98.2,1.6686,4,437.0,21.2,392.04,21.32,13.3
+54,47.6,7.3197,3,469.0,21.1,396.9,14.8,18.9
+271,16.3,4.429,3,223.0,18.6,396.9,6.59,25.2
+423,85.1,2.0218,24,666.0,20.2,2.52,23.29,13.4
+311,52.8,2.6403,4,304.0,18.4,396.9,5.98,22.1
+21,89.2,4.0123,4,307.0,21.0,392.53,13.83,19.6
+453,99.3,2.4527,24,666.0,20.2,375.87,16.74,17.8
+222,77.7,3.2721,8,307.0,17.4,390.39,9.93,27.5

example/FL/linear_regression/hfl_dpsgd.json

+{
+    "party_info": {
+        "task_manager": "127.0.0.1:50050"
+    },
+    "component_params": {
+        "roles": {
+            "server": "Alice",
+            "client": [
+                "Bob",
+                "Charlie"
+            ]
+        },
+        "common_params": {
+            "model": "HFL_linear_regression",
+            "method": "DPSGD",
+            "process": "train",
+            "task_name": "HFL_linear_regression_dpsgd_train",
+            "delta": 1e-3,
+            "noise_multiplier": 2.0,
+            "l2_norm_clip": 1.0,
+            "secure_mode": true,
+            "learning_rate": 1e-1,
+            "alpha": 1e-4,
+            "batch_size": 100,
+            "global_epoch": 10,
+            "local_epoch": 1,
+            "selected_column": null,
+            "id": "id",
+            "label": "y",
+            "print_metrics": true,
+            "metric_path": "data/result/metrics.json"
+        },
+        "role_params": {
+            "Bob": {
+                "data_set": "regression_hfl_train_client1",
+                "model_path": "data/result/Bob_model.pkl"
+            },
+            "Charlie": {
+                "data_set": "regression_hfl_train_client2",
+                "model_path": "data/result/Charlie_model.pkl"
+            },
+            "Alice": {
+                "data_set": "fl_fake_data"
+            }
+        }
+    }
+}
\ No newline at end of file

example/FL/linear_regression/hfl_paillier.json

+{
+    "party_info": {
+        "task_manager": "127.0.0.1:50050"
+    },
+    "component_params": {
+        "roles": {
+            "server": "Alice",
+            "client": [
+                "Bob",
+                "Charlie"
+            ]
+        },
+        "common_params": {
+            "model": "HFL_linear_regression",
+            "method": "Paillier",
+            "process": "train",
+            "task_name": "HFL_linear_regression_paillier_train",
+            "n_length": 2048,
+            "learning_rate": 1e-1,
+            "alpha": 1e-5,
+            "batch_size": 100,
+            "global_epoch": 10,
+            "local_epoch": 1,
+            "selected_column": null,
+            "id": "id",
+            "label": "y",
+            "print_metrics": true,
+            "metric_path": "data/result/metrics.json"
+        },
+        "role_params": {
+            "Bob": {
+                "data_set": "regression_hfl_train_client1",
+                "model_path": "data/result/Bob_model.pkl"
+            },
+            "Charlie": {
+                "data_set": "regression_hfl_train_client2",
+                "model_path": "data/result/Charlie_model.pkl"
+            },
+            "Alice": {
+                "data_set": "fl_fake_data"
+            }
+        }
+    }
+}
\ No newline at end of file

example/FL/linear_regression/hfl_plaintext.json

+{
+    "party_info": {
+        "task_manager": "127.0.0.1:50050"
+    },
+    "component_params": {
+        "roles": {
+            "server": "Alice",
+            "client": [
+                "Bob",
+                "Charlie"
+            ]
+        },
+        "common_params": {
+            "model": "HFL_linear_regression",
+            "method": "Plaintext",
+            "process": "train",
+            "task_name": "HFL_linear_regression_plaintext_train",
+            "learning_rate": 1e-1,
+            "alpha": 1e-5,
+            "batch_size": 100,
+            "global_epoch": 10,
+            "local_epoch": 1,
+            "selected_column": null,
+            "id": "id",
+            "label": "y",
+            "print_metrics": true,
+            "metric_path": "data/result/metrics.json"
+        },
+        "role_params": {
+            "Bob": {
+                "data_set": "regression_hfl_train_client1",
+                "model_path": "data/result/Bob_model.pkl"
+            },
+            "Charlie": {
+                "data_set": "regression_hfl_train_client2",
+                "model_path": "data/result/Charlie_model.pkl"
+            },
+            "Alice": {
+                "data_set": "fl_fake_data"
+            }
+        }
+    }
+}
\ No newline at end of file

example/FL/linear_regression/hfl_predict.json

+{
+    "party_info": {
+        "task_manager": "127.0.0.1:50050"
+    },
+    "component_params": {
+        "roles": {
+            "client": "Bob"
+        },
+        "common_params": {
+            "model": "HFL_linear_regression",
+            "process": "predict",
+            "task_name": "HFL_linear_regression_predict"
+        },
+        "role_params": {
+            "Bob": {
+                "data_set": "regression_hfl_test_client1",
+                "model_path": "data/result/Bob_model.pkl",
+                "predict_path": "data/result/Bob_predict.csv"
+            }
+        }
+    }
+}
\ No newline at end of file

example/FL/linear_regression/vfl_ckks.json

+{
+    "party_info": {
+        "task_manager": "127.0.0.1:50050"
+    },
+    "component_params": {
+        "roles": {
+            "host": "Bob",
+            "guest": [
+                "Charlie"
+            ],
+            "coordinator": "David"
+        },
+        "common_params": {
+            "model": "VFL_linear_regression",
+            "method": "CKKS",
+            "process": "train",
+            "task_name": "VFL_linear_regression_ckks_train",
+            "learning_rate": 1e-1,
+            "alpha": 1e-4,
+            "epoch": 10,
+            "shuffle_seed": 0,
+            "batch_size": 100,
+            "print_metrics": true
+        },
+        "role_params": {
+            "Bob": {
+                "data_set": "regression_vfl_train_host",
+                "selected_column": null,
+                "id": "id",
+                "label": "y",
+                "model_path": "data/result/host_model.pkl",
+                "metric_path": "data/result/metrics.json"
+            },
+            "Charlie": {
+                "data_set": "regression_vfl_train_guest",
+                "selected_column": null,
+                "id": "id",
+                "model_path": "data/result/guest_model.pkl"
+            },
+            "David": {
+                "data_set": "fl_fake_data"
+            }
+        }
+    }
+}
\ No newline at end of file

example/FL/linear_regression/vfl_plaintext.json

+{
+    "party_info": {
+        "task_manager": "127.0.0.1:50050"
+    },
+    "component_params": {
+        "roles": {
+            "host": "Bob",
+            "guest": [
+                "Charlie"
+            ]
+        },
+        "common_params": {
+            "model": "VFL_linear_regression",
+            "method": "Plaintext",
+            "process": "train",
+            "task_name": "VFL_linear_regression_plaintext_train",
+            "learning_rate": 1e-1,
+            "alpha": 1e-4,
+            "epoch": 10,
+            "shuffle_seed": 0,
+            "batch_size": 100,
+            "print_metrics": true
+        },
+        "role_params": {
+            "Bob": {
+                "data_set": "regression_vfl_train_host",
+                "selected_column": null,
+                "id": "id",
+                "label": "y",
+                "model_path": "data/result/host_model.pkl",
+                "metric_path": "data/result/metrics.json"
+            },
+            "Charlie": {
+                "data_set": "regression_vfl_train_guest",
+                "selected_column": null,
+                "id": "id",
+                "model_path": "data/result/guest_model.pkl"
+            }
+        }
+    }
+}
\ No newline at end of file

example/FL/linear_regression/vfl_predict.json

+{
+    "party_info": {
+        "task_manager": "127.0.0.1:50050"
+    },
+    "component_params": {
+        "roles": {
+            "host": "Bob",
+            "guest": [
+                "Charlie"
+            ]
+        },
+        "common_params": {
+            "model": "VFL_linear_regression",
+            "process": "predict",
+            "task_name": "VFL_linear_regression_predict"
+        },
+        "role_params": {
+            "Bob": {
+                "data_set": "regression_vfl_test_host",
+                "model_path": "data/result/host_model.pkl",
+                "predict_path": "data/result/host_predict.csv"
+            },
+            "Charlie": {
+                "data_set": "regression_vfl_test_guest",
+                "model_path": "data/result/guest_model.pkl"
+            }
+        }
+    }
+}
\ No newline at end of file

python/primihub/FL/crypto/ckks.py

+import tenseal as ts
+
+
+class CKKS:
+
+    def __init__(self, context):
+        if isinstance(context, bytes):
+            context = ts.context_from(context)
+        self.context = context
+        self.multiply_depth = context.data.seal_context().first_context_data().chain_index()
+
+    def encrypt_vector(self, vector, context=None):
+        if context:
+            return ts.ckks_vector(context, vector)
+        else:
+            return ts.ckks_vector(self.context, vector)
+    
+    def encrypt_tensor(self, tensor, context=None):
+        if context:
+            return ts.ckks_tensor(context, tensor)
+        else:
+            return ts.ckks_tensor(self.context, tensor)
+        
+    def decrypt(self, ciphertext, secret_key=None):
+        if ciphertext.context().has_secret_key():
+            return ciphertext.decrypt()
+        else:
+            return ciphertext.decrypt(secret_key)
+    
+    def load_vector(self, vector):
+        return ts.ckks_vector_from(self.context, vector)
+
+    def load_tensor(self, tensor):
+        return ts.ckks_tensor_from(self.context, tensor)
\ No newline at end of file

python/primihub/FL/crypto/paillier.py

+class Paillier:
+
+    def __init__(self, public_key, private_key):
+        self.public_key = public_key
+        self.private_key = private_key
+
+    def decrypt_scalar(self, cipher_scalar):
+        return self.private_key.decrypt(cipher_scalar)
+
+    def decrypt_vector(self, cipher_vector):
+        return [self.private_key.decrypt(i) for i in cipher_vector]
+
+    def decrypt_matrix(self, cipher_matrix):
+        return [[self.private_key.decrypt(i) for i in cv] for cv in cipher_matrix]
+    
+    def encrypt_scalar(self, plain_scalar):
+        return self.public_key.encrypt(plain_scalar)
+
+    def encrypt_vector(self, plain_vector):
+        return [self.public_key.encrypt(i) for i in plain_vector]
+
+    def encrypt_matrix(self, plain_matrix):
+        return [[self.private_key.encrypt(i) for i in pv] for pv in plain_matrix]
\ No newline at end of file

python/primihub/FL/linear_regression/base.py

+import numpy as np
+
+
+class LinearRegression:
+
+    # l2 regularization by default, alpha is the penalty parameter
+    def __init__(self, x, learning_rate=0.2, alpha=0.0001):
+
+        self.learning_rate = learning_rate
+        self.alpha = alpha
+
+        self.weight = np.zeros(x.shape[1])
+        self.bias = np.zeros(1)
+
+    def get_theta(self):
+        return np.hstack((self.bias, self.weight))
+
+    def set_theta(self, theta):
+        if not isinstance(theta, np.ndarray):
+            theta = np.array(theta)
+        self.bias = theta[[0]]
+        self.weight = theta[1:]
+
+    def compute_grad(self, x, y):
+        error = self.compute_error(x, y)
+        dw = x.T.dot(error) / x.shape[0] + self.alpha * self.weight
+        db = error.mean(axis=0, keepdims=True)
+        return dw, db
+    
+    def compute_error(self, x, y):
+        z = self.predict(x)
+        return z - y
+
+    def gradient_descent(self, x, y):
+        dw, db = self.compute_grad(x, y)
+        self.weight -= self.learning_rate * dw
+        self.bias -= self.learning_rate * db
+    
+    def fit(self, x, y):
+        self.gradient_descent(x, y)
+
+    def predict(self, x):
+        return x.dot(self.weight) + self.bias
+
+
+class LinearRegression_DPSGD(LinearRegression):
+
+    def __init__(self, x, learning_rate=0.2, alpha=0.0001,
+                 noise_multiplier=1.0, l2_norm_clip=1.0, secure_mode=True):
+        super().__init__(x, learning_rate, alpha)
+        self.noise_multiplier = noise_multiplier
+        self.l2_norm_clip = l2_norm_clip
+        self.secure_mode = secure_mode
+
+    def set_noise_multiplier(self, noise_multiplier):
+        self.noise_multiplier = noise_multiplier
+
+    def set_l2_norm_clip(self, l2_norm_clip):
+        self.l2_norm_clip = l2_norm_clip
+
+    def add_noise(self, x, n=2):
+        # add gaussian noise
+        if self.secure_mode:
+            noise = np.zeros(x.shape)
+            for _ in range(2 * n):
+                noise += np.random.normal(
+                    0, self.l2_norm_clip * self.noise_multiplier, x.shape)
+            noise /= np.sqrt(2 * n)
+        else:
+            noise = np.random.normal(
+                0, self.l2_norm_clip * self.noise_multiplier, x.shape)
+
+        return x + noise
+    
+    def compute_grad(self, x, y):
+        # compute & clip per-example gradient
+        error = self.compute_error(x, y)
+        batch_db = error
+        batch_dw = x * np.expand_dims(error, axis=1)
+        batch_grad_l2_norm = np.sqrt((batch_dw**2).sum(axis=1) + batch_db**2)
+
+        clip = np.maximum(1., batch_grad_l2_norm / self.l2_norm_clip)
+        dw = (batch_dw / np.expand_dims(clip, axis=1)).sum(axis=0)
+        db = (batch_db / clip).sum(axis=0)
+
+        # add gaussian noise
+        dw = self.add_noise(dw) / x.shape[0] + self.alpha * self.weight
+        db = self.add_noise(db) / x.shape[0]
+        return dw, db
+    
+
+class LinearRegression_Paillier(LinearRegression):
+
+    def gradient_descent(self, x, y):
+        error = self.compute_error(x, y)
+        factor = -self.learning_rate / x.shape[0]
+
+        self.weight += (factor * x).T.dot(error) + \
+            (-self.learning_rate * self.alpha) * self.weight
+        self.bias += factor * error.sum(keepdims=True)
\ No newline at end of file

python/primihub/FL/linear_regression/hfl_client.py

+from primihub.FL.utils.net_work import GrpcClient
+from primihub.FL.utils.base import BaseModel
+from primihub.FL.utils.file import check_directory_exist
+from primihub.FL.utils.dataset import read_data,\
+                                      DataLoader,\
+                                      DPDataLoader
+from primihub.utils.logger_util import logger
+from primihub.FL.crypto.paillier import Paillier
+
+import pickle
+import pandas as pd
+import dp_accounting
+from sklearn import metrics
+from .base import LinearRegression,\
+                  LinearRegression_DPSGD,\
+                  LinearRegression_Paillier
+
+
+class LinearRegressionClient(BaseModel):
+    
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        
+    def run(self):
+        process = self.common_params['process']
+        logger.info(f"process: {process}")
+        if process == 'train':
+            self.train()
+        elif process == 'predict':
+            self.predict()
+        else:
+            error_msg = f"Unsupported process: {process}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+    def train(self):
+        # setup communication channels
+        remote_party = self.roles[self.role_params['others_role']]
+        server_channel = GrpcClient(local_party=self.role_params['self_name'],
+                                    remote_party=remote_party,
+                                    node_info=self.node_info,
+                                    task_info=self.task_info)
+
+        # load dataset
+        selected_column = self.common_params['selected_column']
+        id = self.common_params['id']
+        x = read_data(data_info=self.role_params['data'],
+                      selected_column=selected_column,
+                      id=id)
+        label = self.common_params['label']
+        y = x.pop(label).values
+        x = x.values
+
+        # client init
+        method = self.common_params['method']
+        if method == 'Plaintext':
+            client = Plaintext_Client(x,
+                                      self.common_params['learning_rate'],
+                                      self.common_params['alpha'],
+                                      server_channel)
+        elif method == 'DPSGD':
+            client = DPSGD_Client(x,
+                                  self.common_params['learning_rate'],
+                                  self.common_params['alpha'],
+                                  self.common_params['noise_multiplier'],
+                                  self.common_params['l2_norm_clip'],
+                                  self.common_params['secure_mode'],
+                                  server_channel)
+        elif method == 'Paillier':
+            client = Paillier_Client(x, y,
+                                     self.common_params['learning_rate'],
+                                     self.common_params['alpha'],
+                                     server_channel)
+        else:
+            error_msg = f"Unsupported method: {method}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+        # data preprocessing
+        # minmaxscaler
+        data_max = x.max(axis=0)
+        data_min = x.min(axis=0)
+
+        server_channel.send('data_max', data_max)
+        server_channel.send('data_min', data_min)
+
+        data_max = server_channel.recv('data_max')
+        data_min = server_channel.recv('data_min')
+
+        x = (x - data_min) / (data_max - data_min)
+
+        # client training
+        num_examples = client.num_examples
+        batch_size = min(num_examples, self.common_params['batch_size'])
+        if method == 'DPSGD':
+            # DP data loader: Poisson sampling
+            train_dataloader = DPDataLoader(x, y,
+                                            batch_size)
+        else:    
+            train_dataloader = DataLoader(x, y,
+                                          batch_size,
+                                          shuffle=True)
+
+        logger.info("-------- start training --------")
+        global_epoch = self.common_params['global_epoch']
+        for i in range(global_epoch):
+            logger.info(f"-------- global epoch {i+1} / {global_epoch} --------")
+            
+            local_epoch = self.common_params['local_epoch']
+            for j in range(local_epoch):
+                logger.info(f"-------- local epoch {j+1} / {local_epoch} --------")
+                for batch_x, batch_y in train_dataloader:
+                    client.model.fit(batch_x, batch_y)
+
+            client.train()
+
+            # print metrics
+            if self.common_params['print_metrics']:
+                client.print_metrics(x, y)
+        logger.info("-------- finish training --------")
+
+        # send final epsilon when using DPSGD
+        if method == 'DPSGD':
+            delta = self.common_params['delta']
+            steps = global_epoch * local_epoch * num_examples // batch_size
+            eps = client.compute_epsilon(steps, batch_size, delta)
+            server_channel.send("eps", eps)
+            logger.info(f"For delta={delta}, the current epsilon is {eps}")
+        # receive plaintext model when using Paillier
+        elif method == 'Paillier':
+            client.model.set_theta(server_channel.recv("server_model"))
+        
+        # send final metrics
+        client.send_metrics(x, y)
+
+        # save model for prediction
+        modelFile = {
+            "selected_column": selected_column,
+            "id": id,
+            "label": label,
+            "data_max": data_max,
+            "data_min": data_min,
+            "model": client.model
+        }
+        model_path = self.role_params['model_path']
+        check_directory_exist(model_path)
+        logger.info(f"model path: {model_path}")
+        with open(model_path, 'wb') as file_path:
+            pickle.dump(modelFile, file_path)
+
+    def predict(self):
+        # load model for prediction
+        model_path = self.role_params['model_path']
+        logger.info(f"model path: {model_path}")
+        with open(model_path, 'rb') as file_path:
+            modelFile = pickle.load(file_path)
+
+        # load dataset
+        origin_data = read_data(data_info=self.role_params['data'])
+
+        x = origin_data.copy()
+        selected_column = modelFile['selected_column']
+        if selected_column:
+            x = x[selected_column]
+        id = modelFile['id']
+        if id in x.columns:
+            x.pop(id)
+        label = modelFile['label']
+        if label in x.columns:
+            y = x.pop(label).values
+        x = x.values
+
+        # data preprocessing
+        # minmaxscaler
+        data_max = modelFile['data_max']
+        data_min = modelFile['data_min']
+        x = (x - data_min) / (data_max - data_min)
+
+        # test data prediction
+        model = modelFile['model']
+        pred_y = model.predict(x)
+
+        result = pd.DataFrame({
+            'pred_y': pred_y
+        })
+        
+        data_result = pd.concat([origin_data, result], axis=1)
+        predict_path = self.role_params['predict_path']
+        check_directory_exist(predict_path)
+        logger.info(f"predict path: {predict_path}")
+        data_result.to_csv(predict_path, index=False)
+
+
+class Plaintext_Client:
+
+    def __init__(self, x, learning_rate, alpha, server_channel):
+        self.model = LinearRegression(x, learning_rate, alpha)
+        self.param_init(x, server_channel)
+
+    def param_init(self, x, server_channel):
+        self.server_channel = server_channel
+        self.num_examples = x.shape[0]
+        self.send_params()
+
+    def train(self):
+        self.server_channel.send("client_model", self.model.get_theta())
+        self.model.set_theta(self.server_channel.recv("server_model"))
+
+    def send_params(self):
+        self.server_channel.send('num_examples', self.num_examples)
+
+    def send_metrics(self, x, y):
+        y_hat = self.model.predict(x)
+        mse = metrics.mean_squared_error(y, y_hat)
+        mae = metrics.mean_absolute_error(y, y_hat)
+
+        logger.info(f"mse={mse}, mae={mae}")
+
+        self.server_channel.send('mse', mse)
+        self.server_channel.send('mae', mae)
+
+    def print_metrics(self, x, y):
+        self.send_metrics(x, y)
+
+
+class DPSGD_Client(Plaintext_Client):
+
+    def __init__(self, x, learning_rate, alpha,
+                 noise_multiplier, l2_norm_clip, secure_mode,
+                 server_channel):
+        self.model = LinearRegression_DPSGD(x, learning_rate, alpha,
+                                            noise_multiplier, l2_norm_clip, secure_mode)
+        self.param_init(x, server_channel)
+
+    def compute_epsilon(self, steps, batch_size, delta):
+        if self.model.noise_multiplier == 0.0:
+            return float('inf')
+        orders = [1 + x / 10. for x in range(1, 100)] + list(range(12, 64))
+        accountant = dp_accounting.rdp.RdpAccountant(orders)
+
+        sampling_probability = batch_size / self.num_examples
+        event = dp_accounting.SelfComposedDpEvent(
+            dp_accounting.PoissonSampledDpEvent(
+                sampling_probability,
+                dp_accounting.GaussianDpEvent(self.model.noise_multiplier)), steps)
+
+        accountant.compose(event)
+        
+        if delta >= 1. / self.num_examples:
+            logger.error(f"delta {delta} should be set less than 1 / {self.num_examples}")
+        return accountant.get_epsilon(target_delta=delta)
+    
+
+class Paillier_Client(Plaintext_Client, Paillier):
+
+    def __init__(self, x, y, learning_rate, alpha,
+                 server_channel):
+        self.model = LinearRegression_Paillier(x, learning_rate, alpha)
+        self.param_init(x, server_channel)
+
+        self.public_key = server_channel.recv("public_key")
+        self.model.set_theta(self.encrypt_vector(self.model.get_theta()))
+
+    def print_metrics(self, x, y):
+        logger.info('No metrics while using Paillier')
\ No newline at end of file

python/primihub/FL/linear_regression/hfl_server.py

+from primihub.FL.utils.net_work import MultiGrpcClients
+from primihub.FL.utils.base import BaseModel
+from primihub.FL.utils.file import check_directory_exist
+from primihub.utils.logger_util import logger
+from primihub.FL.crypto.paillier import Paillier
+
+import json
+import numpy as np
+from phe import paillier
+
+
+class LinearRegressionServer(BaseModel):
+    
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        
+    def run(self):
+        process = self.common_params['process']
+        logger.info(f"process: {process}")
+        if process == 'train':
+            self.train()
+        else:
+            error_msg = f"Unsupported process: {process}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+    
+    def train(self):
+        # setup communication channels
+        remote_parties = self.roles[self.role_params['others_role']]
+        client_channel = MultiGrpcClients(local_party=self.role_params['self_name'],
+                                          remote_parties=remote_parties,
+                                          node_info=self.node_info,
+                                          task_info=self.task_info)
+
+        # server init
+        method = self.common_params['method']
+        if method == 'Plaintext' or method == 'DPSGD':
+            server = Plaintext_DPSGD_Server(self.common_params['alpha'],
+                                            client_channel)
+        elif method == 'Paillier':
+            server = Paillier_Server(self.common_params['alpha'],
+                                     self.common_params['n_length'],
+                                     client_channel)
+        else:
+            error_msg = f"Unsupported method: {method}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+        # data preprocessing
+        # minmaxscaler
+        data_max = client_channel.recv_all('data_max')
+        data_min = client_channel.recv_all('data_min')
+        
+        data_max = np.array(data_max).max(axis=0)
+        data_min = np.array(data_min).min(axis=0)
+
+        client_channel.send_all('data_max', data_max)
+        client_channel.send_all('data_min', data_min)
+
+        # server training
+        logger.info("-------- start training --------")
+        global_epoch = self.common_params['global_epoch']
+        for i in range(global_epoch):
+            logger.info(f"-------- global epoch {i+1} / {global_epoch} --------")
+            server.train()
+        
+            # print metrics
+            if self.common_params['print_metrics']:
+                server.print_metrics()
+        logger.info("-------- finish training --------")
+
+        # receive final epsilons when using DPSGD
+        if method == 'DPSGD':
+            delta = self.common_params['delta']
+            eps = client_channel.recv_all("eps")
+            logger.info(f"For delta={delta}, the current epsilon is {max(eps)}")
+        # send plaintext model when using Paillier
+        elif method == 'Paillier':
+            server.plaintext_server_model_broadcast()
+
+        # receive final metrics
+        trainMetrics = server.get_metrics()
+        metric_path = self.common_params['metric_path']
+        check_directory_exist(metric_path)
+        logger.info(f"metric path: {metric_path}")
+        with open(metric_path, 'w') as file_path:
+            file_path.write(json.dumps(trainMetrics))
+
+
+class Plaintext_DPSGD_Server:
+
+    def __init__(self, alpha, client_channel):
+        self.alpha = alpha
+        self.client_channel = client_channel
+
+        self.theta = None
+
+        self.num_examples_weights = None
+        self.recv_params()
+
+    def recv_params(self):
+        self.num_examples_weights = self.client_channel.recv_all('num_examples')
+
+    def client_model_aggregate(self):
+        client_models = self.client_channel.recv_all("client_model")
+        
+        self.theta = np.average(client_models,
+                                weights=self.num_examples_weights,
+                                axis=0)
+
+    def server_model_broadcast(self):
+        self.client_channel.send_all("server_model", self.theta)
+
+    def train(self):
+        self.client_model_aggregate()
+        self.server_model_broadcast()
+    
+    def get_scalar_metrics(self, metrics_name):
+        metrics_name = metrics_name.lower()
+        supported_metrics = ['mse', 'mae']
+        if metrics_name not in supported_metrics:
+            error_msg = f"""Unsupported metrics {metrics_name},
+                          use {supported_metrics} instead"""
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+        client_metrics = self.client_channel.recv_all(metrics_name)
+            
+        return np.average(client_metrics,
+                          weights=self.num_examples_weights)
+
+    def get_metrics(self):
+        server_metrics = {}
+
+        mse = self.get_scalar_metrics('mse')
+        server_metrics["train_mse"] = mse
+
+        mae = self.get_scalar_metrics('mae')
+        server_metrics["train_mae"] = mae
+
+        logger.info(f"mse={mse}, mae={mae}")
+
+        return server_metrics
+    
+    def print_metrics(self):
+        self.get_metrics()
+
+
+class Paillier_Server(Plaintext_DPSGD_Server, Paillier):
+    
+    def __init__(self, alpha, n_length, client_channel):
+        Plaintext_DPSGD_Server.__init__(self, alpha, client_channel)
+        self.public_key,\
+        self.private_key = paillier.generate_paillier_keypair(n_length=n_length) 
+        self.public_key_broadcast()
+
+    def public_key_broadcast(self):
+        self.client_channel.send_all("public_key", self.public_key)
+
+    def client_model_aggregate(self):
+        client_models = self.client_channel.recv_all("client_model")
+
+        self.theta = np.mean(client_models, axis=0)
+        self.theta = np.array(self.encrypt_vector(self.decrypt_vector(self.theta)))
+
+    def plaintext_server_model_broadcast(self):
+        self.theta = np.array(self.decrypt_vector(self.theta))
+        self.server_model_broadcast()
+
+    def print_metrics(self):
+        logger.info('No metrics while using Paillier')

python/primihub/FL/linear_regression/vfl_base.py

+import numpy as np
+from primihub.utils.logger_util import logger
+from .base import LinearRegression
+
+
+class LinearRegression_Host_Plaintext(LinearRegression):
+
+    def compute_z(self, x, guest_z):
+        z = x.dot(self.weight) + self.bias
+        z += np.array(guest_z).sum(axis=0)
+        return z
+    
+    def compute_error(self, y, z):
+        return z - y
+    
+    def compute_grad(self, x, error):
+        dw = x.T.dot(error) / x.shape[0] + self.alpha * self.weight
+        db = error.mean(axis=0, keepdims=True)
+        return dw, db
+    
+    def gradient_descent(self, x, error):
+        dw, db = self.compute_grad(x, error)
+        self.weight -= self.learning_rate * dw
+        self.bias -= self.learning_rate * db
+    
+    def fit(self, x, error):
+        self.gradient_descent(x, error)
+
+
+class LinearRegression_Host_CKKS(LinearRegression_Host_Plaintext):
+
+    def compute_enc_z(self, x, guest_z):
+        z = self.weight.mm(x.T) + self.bias
+        z += sum(guest_z)
+        return z
+
+    def gradient_descent(self, x, error):
+        factor = -self.learning_rate / x.shape[0]
+        self.bias += error.sum() * factor
+        self.weight += error.mm(factor * x) \
+            + (-self.learning_rate * self.alpha) * self.weight
+
+
+class LinearRegression_Guest_Plaintext:
+
+    def __init__(self, x, learning_rate=0.2, alpha=0.0001):
+
+        self.learning_rate = learning_rate
+        self.alpha = alpha
+
+        self.weight = np.zeros(x.shape[1])
+    
+    def compute_z(self, x):
+        return x.dot(self.weight)
+    
+    def compute_grad(self, x, error):
+        dw = x.T.dot(error) / x.shape[0] + self.alpha * self.weight
+        return dw
+    
+    def gradient_descent(self, x, error):
+        dw = self.compute_grad(x, error)
+        self.weight -= self.learning_rate * dw
+    
+    def fit(self, x, error):
+        self.gradient_descent(x, error)
+
+
+class LinearRegression_Guest_CKKS(LinearRegression_Guest_Plaintext):
+
+    def compute_enc_z(self, x):
+        return self.weight.mm(x.T)
+    
+    def gradient_descent(self, x, error):
+        factor = -self.learning_rate / x.shape[0]
+        self.weight += error.mm(factor * x) + \
+            (-self.learning_rate * self.alpha) * self.weight

python/primihub/FL/linear_regression/vfl_coordinator.py

+from primihub.FL.utils.net_work import GrpcClient, MultiGrpcClients
+from primihub.FL.utils.base import BaseModel
+from primihub.utils.logger_util import logger
+from primihub.FL.crypto.ckks import CKKS
+
+import math
+import numpy as np
+import tenseal as ts
+
+
+class LinearRegressionCoordinator(BaseModel):
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+    def run(self):
+        process = self.common_params['process']
+        logger.info(f"process: {process}")
+        if process == 'train':
+            self.train()
+        else:
+            error_msg = f"Unsupported process: {process}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+    def train(self):
+        # setup communication channels
+        host_channel = GrpcClient(local_party=self.role_params['self_name'],
+                                  remote_party=self.roles['host'],
+                                  node_info=self.node_info,
+                                  task_info=self.task_info)
+        guest_channel = MultiGrpcClients(local_party=self.role_params['self_name'],
+                                         remote_parties=self.roles['guest'],
+                                         node_info=self.node_info,
+                                         task_info=self.task_info)
+        
+        # coordinator init
+        method = self.common_params['method']
+        if method == 'CKKS':
+            batch_size = host_channel.recv('batch_size')
+            coordinator = CKKSCoordinator(batch_size,
+                                          host_channel,
+                                          guest_channel)
+        else:
+            error_msg = f"Unsupported method: {method}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+        # coordinator training
+        logger.info("-------- start training --------")
+        epoch = self.common_params['epoch']
+        for i in range(epoch):
+            logger.info(f"-------- epoch {i+1} / {epoch} --------")
+            coordinator.train()
+                
+            # print metrics
+            if self.common_params['print_metrics']:
+                coordinator.compute_loss()
+        logger.info("-------- finish training --------")
+
+        # decrypt & send plaintext model
+        coordinator.update_plaintext_model()
+    
+
+class CKKSCoordinator(CKKS):
+
+    def __init__(self, batch_size, host_channel, guest_channel):
+        self.t = 0
+        self.host_channel = host_channel
+        self.guest_channel = guest_channel
+
+        # set CKKS params
+        # use larger poly_mod_degree to support more encrypted multiplications
+        poly_mod_degree = 8192
+        # the least multiplication per iteration of gradient descent
+        # more multiplications lead to larger context size
+        multiply_per_iter = 2
+        self.max_iter = 1
+        multiply_depth = multiply_per_iter * self.max_iter
+        # sum(coeff_mod_bit_sizes) <= max coeff_modulus bit-length
+        fe_bits_scale = 60
+        bits_scale = 49
+        # 60*2 + 49*1*2 = 218 == 218 (for N = 8192 & 128 bit security)
+        coeff_mod_bit_sizes = [fe_bits_scale] + \
+                              [bits_scale] * multiply_depth + \
+                              [fe_bits_scale]
+
+        # create TenSEALContext
+        logger.info('create CKKS TenSEAL context')
+        secret_context = ts.context(ts.SCHEME_TYPE.CKKS,
+                                    poly_modulus_degree=poly_mod_degree,
+                                    coeff_mod_bit_sizes=coeff_mod_bit_sizes)
+        secret_context.global_scale = pow(2, bits_scale)
+        secret_context.generate_galois_keys()
+
+        context = secret_context.copy()
+        context.make_context_public()
+
+        super().__init__(context)
+        self.secret_context = secret_context
+
+        self.send_public_context()
+        self.num_examples = host_channel.recv('num_examples')
+
+        self.iter_per_epoch = math.ceil(self.num_examples / batch_size)
+
+    def send_public_context(self):
+        serialize_context = self.context.serialize()
+        self.host_channel.send("public_context", serialize_context)
+        self.guest_channel.send_all("public_context", serialize_context)
+
+    def recv_model(self):
+        host_weight = self.load_vector(self.host_channel.recv('host_weight'))
+        host_bias = self.load_vector(self.host_channel.recv('host_bias'))
+
+        guest_weight = self.guest_channel.recv_all('guest_weight')
+        guest_weight = [self.load_vector(weight) for weight in guest_weight]
+
+        return host_weight, host_bias, guest_weight
+    
+    def send_model(self, host_weight, host_bias, guest_weight):
+        self.host_channel.send('host_weight', host_weight)
+        self.host_channel.send('host_bias', host_bias)
+        # send n sub-lists to n parties seperately
+        self.guest_channel.send_seperately('guest_weight', guest_weight)
+
+    def decrypt_model(self, host_weight, host_bias, guest_weight):
+        host_weight = self.decrypt(host_weight, self.secret_context.secret_key())
+        host_bias = self.decrypt(host_bias, self.secret_context.secret_key())
+
+        guest_weight = [self.decrypt(weight, self.secret_context.secret_key()) \
+                        for weight in guest_weight]
+        
+        return host_weight, host_bias, guest_weight
+    
+    def encrypt_model(self, host_weight, host_bias, guest_weight):
+        host_weight = self.encrypt_vector(host_weight)
+        host_bias = self.encrypt_vector(host_bias)
+
+        guest_weight = [self.encrypt_vector(weight) for weight in guest_weight]
+        
+        return host_weight, host_bias, guest_weight
+
+    def update_ciphertext_model(self):
+        host_weight, host_bias, guest_weight = self.recv_model()
+        host_weight, host_bias, guest_weight = self.decrypt_model(
+            host_weight, host_bias, guest_weight)
+        host_weight, host_bias, guest_weight = self.encrypt_model(
+            host_weight, host_bias, guest_weight)
+        
+        host_weight = host_weight.serialize()
+        host_bias = host_bias.serialize()
+        guest_weight = [weight.serialize() for weight in guest_weight]
+
+        self.send_model(host_weight, host_bias, guest_weight)
+
+    def update_plaintext_model(self):
+        host_weight, host_bias, guest_weight = self.recv_model()
+        host_weight, host_bias, guest_weight = self.decrypt_model(
+            host_weight, host_bias, guest_weight)
+        
+        # list to numpy ndarrry
+        host_weight = np.array(host_weight)
+        host_bias = np.array(host_bias)
+        guest_weight = [np.array(weight) for weight in guest_weight]
+
+        self.send_model(host_weight, host_bias, guest_weight)
+
+    def train(self):
+        logger.info(f'iteration {self.t} / {self.max_iter}')
+        self.t += self.iter_per_epoch
+        num_dec = self.t // self.max_iter
+        self.t = self.t % self.max_iter
+        if self.t == 0:
+            num_dec -= 1
+            self.t = self.max_iter
+
+        for i in range(num_dec):
+            logger.warning(f'decrypt model #{i+1}')
+            self.update_ciphertext_model()    
+
+    def compute_loss(self):
+        logger.info(f'iteration {self.t} / {self.max_iter}')
+        if self.t >= self.max_iter:
+            self.t = 0
+            logger.warning('decrypt model')
+            self.update_ciphertext_model()
+
+        mse = self.load_vector(self.host_channel.recv('mse'))
+        mse = self.decrypt(mse, self.secret_context.secret_key())[0]
+        mse /= self.num_examples
+        logger.info(f'mse={mse}')
\ No newline at end of file

python/primihub/FL/linear_regression/vfl_guest.py

+from primihub.FL.utils.net_work import GrpcClient
+from primihub.FL.utils.base import BaseModel
+from primihub.FL.utils.file import check_directory_exist
+from primihub.FL.utils.dataset import read_data, DataLoader
+from primihub.utils.logger_util import logger
+from primihub.FL.crypto.ckks import CKKS
+
+import pickle
+from sklearn.preprocessing import StandardScaler
+
+from .vfl_base import LinearRegression_Guest_Plaintext,\
+                      LinearRegression_Guest_CKKS
+
+
+class LinearRegressionGuest(BaseModel):
+    
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        
+    def run(self):
+        process = self.common_params['process']
+        logger.info(f"process: {process}")
+        if process == 'train':
+            self.train()
+        elif process == 'predict':
+            self.predict()
+        else:
+            error_msg = f"Unsupported process: {process}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+    def train(self):
+        # setup communication channels
+        host_channel = GrpcClient(local_party=self.role_params['self_name'],
+                                  remote_party=self.roles['host'],
+                                  node_info=self.node_info,
+                                  task_info=self.task_info)
+        
+        method = self.common_params['method']
+        if method == 'CKKS':
+            coordinator_channel = GrpcClient(local_party=self.role_params['self_name'],
+                                             remote_party=self.roles['coordinator'],
+                                             node_info=self.node_info,
+                                             task_info=self.task_info)
+        
+        # load dataset
+        selected_column = self.role_params['selected_column']
+        id = self.role_params['id']
+        x = read_data(data_info=self.role_params['data'],
+                      selected_column=selected_column,
+                      id=id)
+        x = x.values
+
+        # guest init
+        batch_size = min(x.shape[0], self.common_params['batch_size'])
+        if method == 'Plaintext':
+            guest = Plaintext_Guest(x,
+                                    self.common_params['learning_rate'],
+                                    self.common_params['alpha'],
+                                    host_channel)
+        elif method == 'CKKS':
+            guest = CKKS_Guest(x,
+                               self.common_params['learning_rate'],
+                               self.common_params['alpha'],
+                               host_channel,
+                               coordinator_channel)
+        else:
+            error_msg = f"Unsupported method: {method}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+        # data preprocessing
+        # StandardScaler
+        scaler = StandardScaler()
+        x = scaler.fit_transform(x)
+
+        # guest training
+        train_dataloader = DataLoader(dataset=x,
+                                      label=None,
+                                      batch_size=batch_size,
+                                      shuffle=True,
+                                      seed=self.common_params['shuffle_seed'])
+
+        logger.info("-------- start training --------")
+        epoch = self.common_params['epoch']
+        for i in range(epoch):
+            logger.info(f"-------- epoch {i+1} / {epoch} --------")
+            for batch_x in train_dataloader:
+                guest.train(batch_x)
+        
+            # print metrics
+            if self.common_params['print_metrics']:
+                guest.compute_metrics(x)
+        logger.info("-------- finish training --------")
+
+        # receive plaintext model
+        if method == 'CKKS':
+            guest.update_plaintext_model()
+
+        # compute final metrics
+        guest.compute_final_metrics(x)
+
+        # save model for prediction
+        modelFile = {
+            "selected_column": selected_column,
+            "id": id,
+            "transformer": scaler,
+            "model": guest.model
+        }
+        model_path = self.role_params['model_path']
+        check_directory_exist(model_path)
+        logger.info(f"model path: {model_path}")
+        with open(model_path, 'wb') as file_path:
+            pickle.dump(modelFile, file_path)
+
+    def predict(self):
+        # setup communication channels
+        remote_party = self.roles[self.role_params['others_role']]
+        host_channel = GrpcClient(local_party=self.role_params['self_name'],
+                                  remote_party=remote_party,
+                                  node_info=self.node_info,
+                                  task_info=self.task_info)
+        
+        # load model for prediction
+        model_path = self.role_params['model_path']
+        logger.info(f"model path: {model_path}")
+        with open(model_path, 'rb') as file_path:
+            modelFile = pickle.load(file_path)
+
+        # load dataset
+        x = read_data(data_info=self.role_params['data'])
+
+        selected_column = modelFile['selected_column']
+        if selected_column:
+            x = x[selected_column]
+        id = modelFile['id']
+        if id in x.columns:
+            x.pop(id)
+        x = x.values
+
+        # data preprocessing
+        transformer = modelFile['transformer']
+        x = transformer.transform(x)
+
+        # test data prediction
+        model = modelFile['model']
+        guest_z = model.compute_z(x)
+        host_channel.send('guest_z', guest_z)
+
+
+class Plaintext_Guest:
+
+    def __init__(self, x, learning_rate, alpha,  host_channel):
+        self.model = LinearRegression_Guest_Plaintext(x,
+                                                      learning_rate,
+                                                      alpha)
+        self.host_channel = host_channel
+    
+    def send_z(self, x):
+        guest_z = self.model.compute_z(x)
+        self.host_channel.send('guest_z', guest_z)
+
+    def train(self, x):
+        self.send_z(x)
+
+        error = self.host_channel.recv('error')
+
+        self.model.fit(x, error)
+
+    def compute_metrics(self, x):
+        self.send_z(x)
+
+    def compute_final_metrics(self, x):
+        self.compute_metrics(x)
+            
+
+class CKKS_Guest(Plaintext_Guest, CKKS):
+
+    def __init__(self, x, learning_rate, alpha,
+                 host_channel, coordinator_channel):
+        self.t = 0
+        self.model = LinearRegression_Guest_CKKS(x,
+                                                 learning_rate,
+                                                 alpha)
+        self.host_channel = host_channel
+        self.recv_public_context(coordinator_channel)
+        CKKS.__init__(self, self.context)
+        multiply_per_iter = 2
+        self.max_iter = self.multiply_depth // multiply_per_iter
+        self.encrypt_model()
+    
+    def recv_public_context(self, coordinator_channel):
+        self.coordinator_channel = coordinator_channel
+        self.context = coordinator_channel.recv('public_context')
+
+    def encrypt_model(self):
+        self.model.weight = self.encrypt_vector(self.model.weight)
+
+    def update_ciphertext_model(self):
+        self.coordinator_channel.send('guest_weight',
+                                      self.model.weight.serialize())
+        
+        self.model.weight = self.load_vector(
+            self.coordinator_channel.recv('guest_weight'))
+
+    def update_plaintext_model(self):
+        self.coordinator_channel.send('guest_weight',
+                                      self.model.weight.serialize())
+        self.model.weight = self.coordinator_channel.recv('guest_weight')
+
+    def send_enc_z(self, x):
+        guest_z = self.model.compute_enc_z(x)
+        self.host_channel.send('guest_z', guest_z.serialize())
+    
+    def train(self, x):
+        logger.info(f'iteration {self.t} / {self.max_iter}')
+        if self.t >= self.max_iter:
+            self.t = 0
+            logger.warning(f'decrypt model')
+            self.update_ciphertext_model()
+            
+        self.t += 1
+
+        self.send_enc_z(x)
+
+        error = self.load_vector(self.host_channel.recv('error'))
+
+        self.model.fit(x, error)
+
+    def compute_metrics(self, x):
+        logger.info(f'iteration {self.t} / {self.max_iter}')
+        if self.t >= self.max_iter:
+            self.t = 0
+            logger.warning(f'decrypt model')
+            self.update_ciphertext_model()
+
+        self.send_enc_z(x)
+        logger.info('View metrics at coordinator while using CKKS')
+
+    def compute_final_metrics(self, x):
+        super().compute_metrics(x)
\ No newline at end of file

python/primihub/FL/linear_regression/vfl_host.py

+from primihub.FL.utils.net_work import GrpcClient, MultiGrpcClients
+from primihub.FL.utils.base import BaseModel
+from primihub.FL.utils.file import check_directory_exist
+from primihub.FL.utils.dataset import read_data, DataLoader
+from primihub.utils.logger_util import logger
+from primihub.FL.crypto.ckks import CKKS
+
+import pickle
+import json
+import pandas as pd
+from sklearn import metrics
+from sklearn.preprocessing import StandardScaler
+
+from .vfl_base import LinearRegression_Host_Plaintext,\
+                      LinearRegression_Host_CKKS
+
+
+class LinearRegressionHost(BaseModel):
+    
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        
+    def run(self):
+        process = self.common_params['process']
+        logger.info(f"process: {process}")
+        if process == 'train':
+            self.train()
+        elif process == 'predict':
+            self.predict()
+        else:
+            error_msg = f"Unsupported process: {process}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+    def train(self):
+        # setup communication channels
+        guest_channel = MultiGrpcClients(local_party=self.role_params['self_name'],
+                                         remote_parties=self.roles['guest'],
+                                         node_info=self.node_info,
+                                         task_info=self.task_info)
+        
+        method = self.common_params['method']
+        if method == 'CKKS':
+            coordinator_channel = GrpcClient(local_party=self.role_params['self_name'],
+                                             remote_party=self.roles['coordinator'],
+                                             node_info=self.node_info,
+                                             task_info=self.task_info)
+        
+        # load dataset
+        selected_column = self.role_params['selected_column']
+        id = self.role_params['id']
+        x = read_data(data_info=self.role_params['data'],
+                      selected_column=selected_column,
+                      id=id)
+        label = self.role_params['label']
+        y = x.pop(label).values
+        x = x.values
+
+        # host init
+        batch_size = min(x.shape[0], self.common_params['batch_size'])
+        if method == 'Plaintext':
+            host = Plaintext_Host(x,
+                                  self.common_params['learning_rate'],
+                                  self.common_params['alpha'],
+                                  guest_channel)
+        elif method == 'CKKS':
+            coordinator_channel.send('batch_size', batch_size)
+            host = CKKS_Host(x,
+                             self.common_params['learning_rate'],
+                             self.common_params['alpha'],
+                             guest_channel,
+                             coordinator_channel)
+        else:
+            error_msg = f"Unsupported method: {method}"
+            logger.error(error_msg)
+            raise RuntimeError(error_msg)
+
+        # data preprocessing
+        # StandardScaler
+        scaler = StandardScaler()
+        x = scaler.fit_transform(x)
+        
+        # host training
+        train_dataloader = DataLoader(dataset=x,
+                                      label=y,
+                                      batch_size=batch_size,
+                                      shuffle=True,
+                                      seed=self.common_params['shuffle_seed'])
+        
+        logger.info("-------- start training --------")
+        epoch = self.common_params['epoch']
+        for i in range(epoch):
+            logger.info(f"-------- epoch {i+1} / {epoch} --------")
+            for batch_x, batch_y in train_dataloader:
+                host.train(batch_x, batch_y)
+        
+            # print metrics
+            if self.common_params['print_metrics']:
+                host.compute_metrics(x, y)
+        logger.info("-------- finish training --------")
+
+        # receive plaintext model
+        if method == 'CKKS':
+            host.update_plaintext_model()
+
+        # compute final metrics
+        trainMetrics = host.compute_final_metrics(x, y)
+        metric_path = self.role_params['metric_path']
+        check_directory_exist(metric_path)
+        logger.info(f"metric path: {metric_path}")
+        with open(metric_path, 'w') as file_path:
+            file_path.write(json.dumps(trainMetrics))
+
+        # save model for prediction
+        modelFile = {
+            "selected_column": selected_column,
+            "id": id,
+            "label": label,
+            "transformer": scaler,
+            "model": host.model
+        }
+        model_path = self.role_params['model_path']
+        check_directory_exist(model_path)
+        logger.info(f"model path: {model_path}")
+        with open(model_path, 'wb') as file_path:
+            pickle.dump(modelFile, file_path)
+
+    def predict(self):
+        # setup communication channels
+        remote_parties = self.roles[self.role_params['others_role']]
+        guest_channel = MultiGrpcClients(local_party=self.role_params['self_name'],
+                                         remote_parties=remote_parties,
+                                         node_info=self.node_info,
+                                         task_info=self.task_info)
+        
+        # load model for prediction
+        model_path = self.role_params['model_path']
+        logger.info(f"model path: {model_path}")
+        with open(model_path, 'rb') as file_path:
+            modelFile = pickle.load(file_path)
+
+        # load dataset
+        origin_data = read_data(data_info=self.role_params['data'])
+
+        x = origin_data.copy()
+        selected_column = modelFile['selected_column']
+        if selected_column:
+            x = x[selected_column]
+        id = modelFile['id']
+        if id in x.columns:
+            x.pop(id)
+        label = modelFile['label']
+        if label in x.columns:
+            y = x.pop(label).values
+        x = x.values
+
+        # data preprocessing
+        transformer = modelFile['transformer']
+        x = transformer.transform(x)
+
+        # test data prediction
+        model = modelFile['model']
+        guest_z = guest_channel.recv_all('guest_z')
+        z = model.compute_z(x, guest_z)
+
+        result = pd.DataFrame({
+            'pred_y': z
+        })
+        
+        data_result = pd.concat([origin_data, result], axis=1)
+        predict_path = self.role_params['predict_path']
+        check_directory_exist(predict_path)
+        logger.info(f"predict path: {predict_path}")
+        data_result.to_csv(predict_path, index=False)
+
+
+class Plaintext_Host:
+
+    def __init__(self, x, learning_rate, alpha, guest_channel):
+        self.model = LinearRegression_Host_Plaintext(x,
+                                                     learning_rate,
+                                                     alpha)
+        self.guest_channel = guest_channel
+
+    def compute_z(self, x):
+        guest_z = self.guest_channel.recv_all('guest_z')
+        return self.model.compute_z(x, guest_z)
+        
+    def train(self, x, y):
+        z = self.compute_z(x)
+        
+        error = self.model.compute_error(y, z)
+        self.guest_channel.send_all('error', error)
+
+        self.model.fit(x, error)
+
+    def compute_metrics(self, x, y):
+        z = self.compute_z(x)
+        
+        mse = metrics.mean_squared_error(y, z)
+        mae = metrics.mean_absolute_error(y, z)
+
+        logger.info(f"mse={mse}, mae={mae}")
+        return {
+            'train_mse': mse,
+            'train_mae': mae
+        }
+    
+    def compute_final_metrics(self, x, y):
+        return self.compute_metrics(x, y)
+
+
+class CKKS_Host(Plaintext_Host, CKKS):
+
+    def __init__(self, x, learning_rate, alpha,
+                 guest_channel, coordinator_channel):
+        self.t = 0
+        self.model = LinearRegression_Host_CKKS(x,
+                                                learning_rate,
+                                                alpha)
+        self.guest_channel = guest_channel
+        self.recv_public_context(coordinator_channel)
+        coordinator_channel.send('num_examples', x.shape[0])
+        CKKS.__init__(self, self.context)
+        multiply_per_iter = 2
+        self.max_iter = self.multiply_depth // multiply_per_iter
+        self.encrypt_model()
+
+    def recv_public_context(self, coordinator_channel):
+        self.coordinator_channel = coordinator_channel
+        self.context = coordinator_channel.recv('public_context')
+
+    def encrypt_model(self):
+        self.model.weight = self.encrypt_vector(self.model.weight)
+        self.model.bias = self.encrypt_vector(self.model.bias)
+
+    def update_ciphertext_model(self):
+        self.coordinator_channel.send('host_weight',
+                                      self.model.weight.serialize())
+        self.coordinator_channel.send('host_bias',
+                                      self.model.bias.serialize())
+        
+        self.model.weight = self.load_vector(
+            self.coordinator_channel.recv('host_weight'))
+        self.model.bias = self.load_vector(
+            self.coordinator_channel.recv('host_bias'))
+        
+    def update_plaintext_model(self):
+        self.coordinator_channel.send('host_weight',
+                                      self.model.weight.serialize())
+        self.coordinator_channel.send('host_bias',
+                                      self.model.bias.serialize())
+        self.model.weight = self.coordinator_channel.recv('host_weight')
+        self.model.bias = self.coordinator_channel.recv('host_bias')
+
+    def compute_enc_z(self, x):
+        guest_z = self.guest_channel.recv_all('guest_z')
+        guest_z = [self.load_vector(z) for z in guest_z]
+        return self.model.compute_enc_z(x, guest_z)
+    
+    def train(self, x, y):
+        logger.info(f'iteration {self.t} / {self.max_iter}')
+        if self.t >= self.max_iter:
+            self.t = 0
+            logger.warning(f'decrypt model')
+            self.update_ciphertext_model()
+            
+        self.t += 1
+
+        z = self.compute_enc_z(x)
+        
+        error = self.model.compute_error(y, z)
+        self.guest_channel.send_all('error', error.serialize())
+
+        self.model.fit(x, error)
+
+    def compute_metrics(self, x, y):
+        logger.info(f'iteration {self.t} / {self.max_iter}')
+        if self.t >= self.max_iter:
+            self.t = 0
+            logger.warning(f'decrypt model')
+            self.update_ciphertext_model()
+
+        z = self.compute_enc_z(x)
+
+        mse = ((z - y) ** 2).sum()
+        self.coordinator_channel.send('mse', mse.serialize())
+        logger.info('View metrics at coordinator while using CKKS')
+    
+    def compute_final_metrics(self, x, y):
+        return super().compute_metrics(x, y)
\ No newline at end of file

python/primihub/FL/logistic_regression/base.py

@@ -122,7 +122,7 @@ class LogisticRegression_DPSGD(LogisticRegression):
 
         return x + noise
     
-    def compute_grad(self, x, y): 
+    def compute_grad(self, x, y):
         # compute & clip per-example gradient
         if self.multiclass:
             error = self.predict_prob(x)
@@ -149,32 +149,7 @@ class LogisticRegression_DPSGD(LogisticRegression):
         # add gaussian noise
         dw = self.add_noise(dw) / x.shape[0] + self.alpha * self.weight
         db = self.add_noise(db) / x.shape[0]
-        return dw, db 
-
-
-class Paillier:
-
-    def __init__(self, public_key, private_key):
-        self.public_key = public_key
-        self.private_key = private_key
-
-    def decrypt_scalar(self, cipher_scalar):
-        return self.private_key.decrypt(cipher_scalar)
-
-    def decrypt_vector(self, cipher_vector):
-        return [self.private_key.decrypt(i) for i in cipher_vector]
-
-    def decrypt_matrix(self, cipher_matrix):
-        return [[self.private_key.decrypt(i) for i in cv] for cv in cipher_matrix]
-    
-    def encrypt_scalar(self, plain_scalar):
-        return self.public_key.encrypt(plain_scalar)
-
-    def encrypt_vector(self, plain_vector):
-        return [self.public_key.encrypt(i) for i in plain_vector]
-
-    def encrypt_matrix(self, plain_matrix):
-        return [[self.private_key.encrypt(i) for i in pv] for pv in plain_matrix]
+        return dw, db
 
 
 class LogisticRegression_Paillier(LogisticRegression):

python/primihub/FL/logistic_regression/hfl_client.py

@@ -5,6 +5,7 @@ from primihub.FL.utils.dataset import read_data,\
                                       DataLoader,\
                                       DPDataLoader
 from primihub.utils.logger_util import logger
+from primihub.FL.crypto.paillier import Paillier
 
 import pickle
 import pandas as pd
@@ -15,8 +16,7 @@ from primihub.FL.metrics.hfl_metrics import ks_from_fpr_tpr,\
                                             auc_from_fpr_tpr
 from .base import LogisticRegression,\
                   LogisticRegression_DPSGD,\
-                  LogisticRegression_Paillier,\
-                  Paillier
+                  LogisticRegression_Paillier
 
 
 class LogisticRegressionClient(BaseModel):

python/primihub/FL/logistic_regression/hfl_server.py

@@ -2,6 +2,7 @@ from primihub.FL.utils.net_work import MultiGrpcClients
 from primihub.FL.utils.base import BaseModel
 from primihub.FL.utils.file import check_directory_exist
 from primihub.utils.logger_util import logger
+from primihub.FL.crypto.paillier import Paillier
 
 import json
 import numpy as np
@@ -9,7 +10,6 @@ from phe import paillier
 from primihub.FL.metrics.hfl_metrics import roc_vertical_avg,\
                                             ks_from_fpr_tpr,\
                                             auc_from_fpr_tpr
-from .base import Paillier
 
 
 class LogisticRegressionServer(BaseModel):

python/primihub/FL/logistic_regression/vfl_coordinator.py

 from primihub.FL.utils.net_work import GrpcClient, MultiGrpcClients
 from primihub.FL.utils.base import BaseModel
 from primihub.utils.logger_util import logger
+from primihub.FL.crypto.ckks import CKKS
 
 import math
 import numpy as np
@@ -59,39 +60,6 @@ class LogisticRegressionCoordinator(BaseModel):
 
         # decrypt & send plaintext model
         coordinator.update_plaintext_model()
-
-
-class CKKS:
-
-    def __init__(self, context):
-        if isinstance(context, bytes):
-            context = ts.context_from(context)
-        self.context = context
-        self.multiply_depth = context.data.seal_context().first_context_data().chain_index()
-
-    def encrypt_vector(self, vector, context=None):
-        if context:
-            return ts.ckks_vector(context, vector)
-        else:
-            return ts.ckks_vector(self.context, vector)
-    
-    def encrypt_tensor(self, tensor, context=None):
-        if context:
-            return ts.ckks_tensor(context, tensor)
-        else:
-            return ts.ckks_tensor(self.context, tensor)
-        
-    def decrypt(self, ciphertext, secret_key=None):
-        if ciphertext.context().has_secret_key():
-            return ciphertext.decrypt()
-        else:
-            return ciphertext.decrypt(secret_key)
-    
-    def load_vector(self, vector):
-        return ts.ckks_vector_from(self.context, vector)
-
-    def load_tensor(self, tensor):
-        return ts.ckks_tensor_from(self.context, tensor)
     
 
 class CKKSCoordinator(CKKS):

python/primihub/FL/logistic_regression/vfl_guest.py

@@ -3,13 +3,13 @@ from primihub.FL.utils.base import BaseModel
 from primihub.FL.utils.file import check_directory_exist
 from primihub.FL.utils.dataset import read_data, DataLoader
 from primihub.utils.logger_util import logger
+from primihub.FL.crypto.ckks import CKKS
 
 import pickle
 from sklearn.preprocessing import StandardScaler
 
 from .vfl_base import LogisticRegression_Guest_Plaintext,\
                       LogisticRegression_Guest_CKKS
-from .vfl_coordinator import CKKS
 
 
 class LogisticRegressionGuest(BaseModel):

python/primihub/FL/logistic_regression/vfl_host.py

@@ -3,6 +3,7 @@ from primihub.FL.utils.base import BaseModel
 from primihub.FL.utils.file import check_directory_exist
 from primihub.FL.utils.dataset import read_data, DataLoader
 from primihub.utils.logger_util import logger
+from primihub.FL.crypto.ckks import CKKS
 
 import pickle
 import json
@@ -15,7 +16,6 @@ from sklearn.preprocessing import StandardScaler
 
 from .vfl_base import LogisticRegression_Host_Plaintext,\
                       LogisticRegression_Host_CKKS
-from .vfl_coordinator import CKKS
 
 
 class LogisticRegressionHost(BaseModel):
@@ -360,7 +360,7 @@ class CKKS_Host(Plaintext_Host, CKKS):
         regular_loss = self.compute_enc_regular_loss()
         loss = self.model.loss(y, z, regular_loss)
         self.coordinator_channel.send('loss', loss.serialize())
-        logger.info('View metrics at coordinator while using Paillier')
+        logger.info('View metrics at coordinator while using CKKS')
     
     def compute_final_metrics(self, x, y):
         return super().compute_metrics(x, y)
\ No newline at end of file

python/primihub/FL/model_map.json

@@ -12,6 +12,15 @@
         "host": "primihub.FL.logistic_regression.vfl_host.LogisticRegressionHost",
         "coordinator": "primihub.FL.logistic_regression.vfl_coordinator.LogisticRegressionCoordinator"
     },
+    "HFL_linear_regression": {
+        "client": "primihub.FL.linear_regression.hfl_client.LinearRegressionClient",
+        "server": "primihub.FL.linear_regression.hfl_server.LinearRegressionServer"
+    },
+    "VFL_linear_regression": {
+        "guest": "primihub.FL.linear_regression.vfl_guest.LinearRegressionGuest",
+        "host": "primihub.FL.linear_regression.vfl_host.LinearRegressionHost",
+        "coordinator": "primihub.FL.linear_regression.vfl_coordinator.LinearRegressionCoordinator"
+    },
     "HFL_neural_network": {
         "client": "primihub.FL.neural_network.hfl_client.NeuralNetworkClient",
         "server": "primihub.FL.neural_network.hfl_server.NeuralNetworkServer"