Table 3 Performance comparison with current state-of-the-art methods on dataset-a

MGCN

0.850

0.707

0.663

0.552

0.634

0.734

3.349

1.266

1.113

[39]

(0.064)

(0.016)

(0.009)

(0.018)

(0.042)

(0.030)

(0.097)

(0.147)

(0.041)

SchNet

0.847

0.767

0.679

0.545

0.717

0.750

3.215

1.045

0.909

[51]

(0.024)

(0.025)

(0.021)

(0.038)

(0.042)

(0.033)

(0.755)

(0.064)

(0.098)

Weave

0.837

0.741

0.678

0.543

0.823

0.791

2.398

1.158

0.813

[29]

(0.065)

(0.044)

(0.024)

(0.034)

(0.023)

(0.008)

(0.250)

(0.055)

(0.042)

GraphConv

0.877

0.772

0.650

0.593

0.845

0.854

2.900

1.068

0.712

[31]

(0.036)

(0.041)

(0.025)

(0.035)

(0.051)

(0.011)

(0.135)

(0.050)

(0.049)

HU.et.al

0.915

0.811

0.714

0.614

0.762

0.851

-

-

-

[22]

(0.040)

(0.015)

(0.019)

(0.006)

(0.058)

(0.027)

AttentiveFP

0.908

0.807

0.579

0.605

0.933

0.863

2.030

0.853

0.650

[62]

(0.050)

(0.020)

(0.001)

(0.060)

(0.020)

(0.015)

(0.420)

(0.060)

(0.030)

MPNN

0.913

0.808

0.691

0.595

0.879

0.815

2.185

1.167

0.672

[a]^a

[19]

(0.041)

(0.024)

(0.013)

(0.030)

(0.054)

(0.044)

(0.952)

(0.430)

(0.051)

DMPNN

0.919

0.826

0.718

0.632

0.897

0.852

2.177

0.980

0.653

[63]

(0.030)

(0.023)

(0.011)

(0.023)

(0.040)

(0.053)

(0.914)

(0.258)

(0.046)

MPG

0.935

0.805

0.712

0.628

0.915

0.839

2.286

0.908

0.637

[35]

(0.012)

(0.015)

(0.010)

(0.020)

(0.023)

(0.052)

(0.389)

(0.105)

(0.044)

GROVER

0.940

0.831

0.737

0.658

0.944

0.894

1.544

0.831

0.560

[50]

(0.019)

(0.025)

(0.010)

(0.023)

(0.021)

(0.028)

(0.397)

(0.120)

(0.035)

KANO

0.939

0.808

0.723

0.626

0.938

0.852

1.938

0.840

0.596

[16]

(0.016)

(0.018)

(0.008)

(0.009)

(0.022)

(0.034)

(0.288)

(0.150)

(0.036)

Our model

0.956

0.822

0.749

0.626

0.921

0.849

1.940

0.746

0.597

(0.018)

(0.029)

(0.001)

(0.007)

(0.035)

(0.028)

(0.312)

(0.093)

(0.027)