Skip to main content

Table 4 Abs(Bias (%)) for various models on the test set

From: Wheat physiology predictor: predicting physiological traits in wheat from hyperspectral reflectance measurements using deep learning

Model

LMA

Narea

SPAD

Nmass

Vcmax

Vcmax25

J

A

gs

Vcmax25/Narea

Mean

Multitask 1DCNN

0.754

0.737

1.150

0.556

1.502

0.614

0.866

0.817

2.683

0.905

1.058

Multitask 1DCNN, No Spectral Trim

0.704

1.105

0.574

1.001

2.619

2.359

1.878

2.390

3.842

0.678

1.715

Singletask 1DCNN

2.307

1.242

0.947

2.510

3.000

0.779

1.081

3.735

6.968

0.780

2.335

Singletask 1DCNN, No Spectral Trim

1.183

0.465

0.935

2.466

1.809

1.808

1.652

2.366

4.559

0.738

1.798

MLP

1.878

0.596

0.921

3.120

1.765

1.509

1.692

2.232

2.752

0.582

1.705

LSTM

0.927

1.393

1.115

0.809

2.463

0.837

1.319

1.687

2.309

2.775

1.563

PLSR

0.322

0.291

0.725

0.934

3.664

0.977

0.475

1.816

5.049

1.389

1.564

XGBoost

0.477

0.007

1.198

2.251

3.421

1.647

0.333

0.079

2.463

3.654

1.553

Ensemble

0.912

0.328

0.941

1.060

2.630

0.229

0.588

2.085

4.899

0.416

1.409

  1. The results reported are the mean of three runs with different random seeds
Back to article page

Plant Methods

ISSN: 1746-4811