Table 4 Model performance comparison across different datasets showing training/test/validation accuracies and computational efficiency. Missing values are indicated with “–”. All accuracy scores range [0,1], with higher values indicating better performance. CPU times in seconds.

Iris

ASM

1.0000

1.0000

0.9231

0.0013

ASRM

1.0000

1.0000

0.9231

0.0014

GD

0.3929

0.4445

0.3810

4.0429

Digits

ASM

0.9638

0.9463

0.9683

0.0885

ASRM

0.9638

0.9463

0.9683

0.0527

GD

0.7990

0.7926

0.7857

4.7482

MNIST

ASM

0.8711

0.8718

0.8692

0.8631

ASRM

0.8711

0.8718

0.8692

2.8285

GD

0.9333

0.9262

0.9266

55.9809

Breast Cancer

ASM

0.9693

0.9474

0.9500

0.0030

ASRM

0.9693

0.9474

0.9500

0.0030

GD

0.6289

0.6316

0.6125

4.3957

Wine

ASM

1.0000

1.0000

0.9231

0.0030

ASRM

1.0000

1.0000

0.9231

0.0021

GD

0.3839

0.3889

0.4400

3.2338

Fashion MNIST

ASM

0.8331

0.8139

0.8218

0.8109

ASRM

0.8332

0.8147

0.8234

2.8411

GD

0.7389

0.7327

0.7365

59.6885

SENTIMENT

ASM

0.8433

0.8257

0.8228

0.1395

ASRM

0.8429

0.8271

0.8209

0.7191

GD

0.8462

0.8233

0.8205

4.8094

IFLYTEK

ASM

0.7228

–

0.5063

0.5036

ASRM

0.7230

–

0.5068

1.3791

GD

0.9220

–

0.4475

42.7414

TNEWS

ASM

0.5100

–

0.4826

0.4174

ASRM

0.5102

–

0.4830

2.2367

GD

0.5315

–

0.4683

11.8202

SHOPPING

ASM

0.7878

–

0.7689

0.4483

ASRM

0.7880

–

0.7688

2.0602

GD

0.8285

–

0.7794

56.5898