Comparison of Machine Learning Regression Algorithms for Cotton Leaf Area Index Retrieval Using Sentinel-2 Spectral Bands

Leaf area index (LAI) is a crucial crop biophysical parameter that has been widely used in a variety of fields. Five state-of-the-art machine learning regression algorithms (MLRAs), namely, artificial neural network (ANN), support vector regression (SVR), Gaussian process regression (GPR), random forest (RF) and gradient boosting regression tree (GBRT), have been used in the retrieval of cotton LAI with Sentinel-2 spectral bands. The performances of the five machine learning models are compared for better applications of MLRAs in remote sensing, since challenging problems remain in the selection of MLRAs for crop LAI retrieval, as well as the decision as to the optimal number for the training sample size and spectral bands to different MLRAs. A comprehensive evaluation was employed with respect to model accuracy, computational efficiency, sensitivity to training sample size and sensitivity to spectral bands. We conducted the comparison of five MLRAs in an agricultural area of Northwest China over three cotton seasons with the corresponding field campaigns for modeling and validation. Results show that the GBRT model outperforms the other models with respect to model accuracy in average ( R 2 ¯ = 0.854, R M S E ¯ = 0.674 and M A E ¯ = 0.456). SVR achieves the best performance in computational efficiency, which means it is fast to train, and to validate that it has great potentials to deliver near-real-time operational products for crop management. As for sensitivity to training sample size, GBRT behaves as the most robust model, and provides the best model accuracy on the average among the variations of training sample size, compared with other models ( R 2 ¯ = 0.884, R M S E ¯ = 0.615 and M A E ¯ = 0.452). Spectral bands sensitivity analysis with dCor (distance correlation), combined with the backward elimination approach, indicates that SVR, GPR and RF provide relatively robust performance to the spectral bands, while ANN outperforms the other models in terms of model accuracy on the average among the reduction of spectral bands ( R 2 ¯ = 0.881, R M S E ¯ = 0.625 and M A E ¯ = 0.480). A comprehensive evaluation indicates that GBRT is an appealing alternative for cotton LAI retrieval, except for its computational efficiency. Despite the different performance of the ML models, all models exhibited considerable potential for cotton LAI retrieval, which could offer accurate crop parameters information timely and accurately for crop fields management and agricultural production decisions.

Download Full-text

Impact of Training Sample Size on the Effects of Regularization in a Convolutional Neural Network-based Dental X-ray Artifact Prediction Model

Journal of Undergraduate Life Sciences ◽

10.33137/juls.v14i1.35883 ◽

2020 ◽

Vol 14 (1) ◽

pp. 5

Author(s):

Adam Adli ◽

Pascal Tyrrell

Keyword(s):

Neural Network ◽

Machine Learning ◽

Convolutional Neural Network ◽

Sample Size ◽

Training Sample ◽

Training Data ◽

Classification Model ◽

Sample Sizes ◽

X Ray ◽

Training Sample Size

Introduction: Advances in computers have allowed for the practical application of increasingly advanced machine learning models to aid healthcare providers with diagnosis and inspection of medical images. Often, a lack of training data and computation time can be a limiting factor in the development of an accurate machine learning model in the domain of medical imaging. As a possible solution, this study investigated whether L2 regularization moderate s the overfitting that occurs as a result of small training sample sizes.Methods: This study employed transfer learning experiments on a dental x-ray binary classification model to explore L2 regularization with respect to training sample size in five common convolutional neural network architectures. Model testing performance was investigated and technical implementation details including computation times and hardware considerations as well as performance factors and practical feasibility were described.Results: The experimental results showed a trend that smaller training sample sizes benefitted more from regularization than larger training sample sizes. Further, the results showed that applying L2 regularization did not apply significant computational overhead and that the extra rounds of training L2 regularization were feasible when training sample sizes are relatively small.Conclusion: Overall, this study found that there is a window of opportunity in which the benefits of employing regularization can be most cost-effective relative to training sample size. It is recommended that training sample size should be carefully considered when forming expectations of achievable generalizability improvements that result from investing computational resources into model regularization.

Download Full-text

The impact of training sample size on deep learning-based organ auto-segmentation for head-and-neck patients

Physics in Medicine and Biology ◽

10.1088/1361-6560/ac2206 ◽

2021 ◽

Vol 66 (18) ◽

pp. 185012

Author(s):

Yingtao Fang ◽

Jiazhou Wang ◽

Xiaomin Ou ◽

Hongmei Ying ◽

Chaosu Hu ◽

...

Keyword(s):

Deep Learning ◽

Sample Size ◽

Head And Neck ◽

Training Sample ◽

Training Sample Size ◽

The Impact

Download Full-text

Sensitivity of hyperspectral classification algorithms to training sample size

2009 First Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing ◽

10.1109/whispers.2009.5288983 ◽

2009 ◽

Cited By ~ 11

Author(s):

M.A. Lee ◽

S. Prasad ◽

L.M. Bruce ◽

T.R. West ◽

D. Reynolds ◽

...

Keyword(s):

Sample Size ◽

Training Sample ◽

Classification Algorithms ◽

Training Sample Size ◽

Hyperspectral Classification

Download Full-text

The Effect of Training Sample Size on the Prediction of White Matter Hyperintensity Volume in a Healthy Population Using BIANCA

Frontiers in Aging Neuroscience ◽

10.3389/fnagi.2021.720636 ◽

2022 ◽

Vol 13 ◽

Author(s):

Niklas Wulms ◽

Lea Redmann ◽

Christine Herpertz ◽

Nadine Bonberg ◽

Klaus Berger ◽

...

Keyword(s):

White Matter ◽

Sample Size ◽

Mean Absolute Error ◽

External Validation ◽

Similarity Index ◽

Absolute Error ◽

Training Sample ◽

White Matter Hyperintensity ◽

Training Sample Size ◽

The Difference

Introduction: White matter hyperintensities of presumed vascular origin (WMH) are an important magnetic resonance imaging marker of cerebral small vessel disease and are associated with cognitive decline, stroke, and mortality. Their relevance in healthy individuals, however, is less clear. This is partly due to the methodological challenge of accurately measuring rare and small WMH with automated segmentation programs. In this study, we tested whether WMH volumetry with FMRIB software library v6.0 (FSL; https://fsl.fmrib.ox.ac.uk/fsl/fslwiki) Brain Intensity AbNormality Classification Algorithm (BIANCA), a customizable and trainable algorithm that quantifies WMH volume based on individual data training sets, can be optimized for a normal aging population.Methods: We evaluated the effect of varying training sample sizes on the accuracy and the robustness of the predicted white matter hyperintensity volume in a population (n = 201) with a low prevalence of confluent WMH and a substantial proportion of participants without WMH. BIANCA was trained with seven different sample sizes between 10 and 40 with increments of 5. For each sample size, 100 random samples of T1w and FLAIR images were drawn and trained with manually delineated masks. For validation, we defined an internal and external validation set and compared the mean absolute error, resulting from the difference between manually delineated and predicted WMH volumes for each set. For spatial overlap, we calculated the Dice similarity index (SI) for the external validation cohort.Results: The study population had a median WMH volume of 0.34 ml (IQR of 1.6 ml) and included n = 28 (18%) participants without any WMH. The mean absolute error of the difference between BIANCA prediction and manually delineated masks was minimized and became more robust with an increasing number of training participants. The lowest mean absolute error of 0.05 ml (SD of 0.24 ml) was identified in the external validation set with a training sample size of 35. Compared to the volumetric overlap, the spatial overlap was poor with an average Dice similarity index of 0.14 (SD 0.16) in the external cohort, driven by subjects with very low lesion volumes.Discussion: We found that the performance of BIANCA, particularly the robustness of predictions, could be optimized for use in populations with a low WMH load by enlargement of the training sample size. Further work is needed to evaluate and potentially improve the prediction accuracy for low lesion volumes. These findings are important for current and future population-based studies with the majority of participants being normal aging people.

Download Full-text

On the relationship between training sample size and data dimensionality: Monte Carlo analysis of broadband multi-temporal classification

Remote Sensing of Environment ◽

10.1016/j.rse.2005.08.011 ◽

2005 ◽

Vol 98 (4) ◽

pp. 468-480 ◽

Cited By ~ 105

Author(s):

T VANNIEL ◽

T MCVICAR ◽

B DATT

Keyword(s):

Monte Carlo ◽

Sample Size ◽

Monte Carlo Analysis ◽

Training Sample ◽

Training Sample Size ◽

Multi Temporal ◽

The Relationship

Download Full-text

Breast Cancer Diagnosis in Digital Breast Tomosynthesis: Effects of Training Sample Size on Multi-Stage Transfer Learning Using Deep Neural Nets

IEEE Transactions on Medical Imaging ◽

10.1109/tmi.2018.2870343 ◽

2019 ◽

Vol 38 (3) ◽

pp. 686-696 ◽

Cited By ~ 32

Author(s):

Ravi K. Samala ◽

Heang-Ping Chan ◽

Lubomir Hadjiiski ◽

Mark A. Helvie ◽

Caleb D. Richter ◽

...

Keyword(s):

Breast Cancer ◽

Sample Size ◽

Cancer Diagnosis ◽

Digital Breast Tomosynthesis ◽

Breast Cancer Diagnosis ◽

Training Sample ◽

Neural Nets ◽

Breast Tomosynthesis ◽

Training Sample Size ◽

Multi Stage

Download Full-text

The Effect of Training Sample Size on Performance of Mass Detection

Digital Mammography - Lecture Notes in Computer Science ◽

10.1007/978-3-540-70538-3_48 ◽

2008 ◽

pp. 343-349 ◽

Cited By ~ 2

Author(s):

Michiel Kallenberg ◽

Nico Karssemeijer

Keyword(s):

Sample Size ◽

Training Sample ◽

Mass Detection ◽

Training Sample Size

Download Full-text

The Value of Sentinel-2 Spectral Bands for the Assessment of Winter Wheat Growth and Development

Remote Sensing ◽

10.3390/rs11172050 ◽

2019 ◽

Vol 11 (17) ◽

pp. 2050 ◽

Cited By ~ 8

Author(s):

Andrew Revill ◽

Anna Florence ◽

Alasdair MacArthur ◽

Stephen Hoad ◽

Robert Rees ◽

...

Keyword(s):

Winter Wheat ◽

Chlorophyll Content ◽

Management Practices ◽

Near Infrared ◽

Spectral Characteristics ◽

Crop Management ◽

Area Index ◽

Spectral Bands ◽

The Mean ◽

Sentinel 2

Leaf Area Index (LAI) and chlorophyll content are strongly related to plant development and productivity. Spatial and temporal estimates of these variables are essential for efficient and precise crop management. The availability of open-access data from the European Space Agency’s (ESA) Sentinel-2 satellite—delivering global coverage with an average 5-day revisit frequency at a spatial resolution of up to 10 metres—could provide estimates of these variables at unprecedented (i.e., sub-field) resolution. Using synthetic data, past research has demonstrated the potential of Sentinel-2 for estimating crop variables. Nonetheless, research involving a robust analysis of the Sentinel-2 bands for supporting agricultural applications is limited. We evaluated the potential of Sentinel-2 data for retrieving winter wheat LAI, leaf chlorophyll content (LCC) and canopy chlorophyll content (CCC). In coordination with destructive and non-destructive ground measurements, we acquired multispectral data from an Unmanned Aerial Vehicle (UAV)-mounted sensor measuring key Sentinel-2 spectral bands (443 to 865 nm). We applied Gaussian processes regression (GPR) machine learning to determine the most informative Sentinel-2 bands for retrieving each of the variables. We further evaluated the GPR model performance when propagating observation uncertainty. When applying the best-performing GPR models without propagating uncertainty, the retrievals had a high agreement with ground measurements—the mean R2 and normalised root-mean-square error (NRMSE) were 0.89 and 8.8%, respectively. When propagating uncertainty, the mean R2 and NRMSE were 0.82 and 11.9%, respectively. When accounting for measurement uncertainty in the estimation of LAI and CCC, the number of most informative Sentinel-2 bands was reduced from four to only two—the red-edge (705 nm) and near-infrared (865 nm) bands. This research demonstrates the value of the Sentinel-2 spectral characteristics for retrieving critical variables that can support more sustainable crop management practices.

Download Full-text

A Comparison of Hybrid Machine Learning Algorithms for the Retrieval of Wheat Biophysical Variables from Sentinel-2

Remote Sensing ◽

10.3390/rs11050481 ◽

2019 ◽

Vol 11 (5) ◽

pp. 481 ◽

Cited By ~ 26

Author(s):

Deepak Upreti ◽

Wenjiang Huang ◽

Weiping Kong ◽

Simone Pascucci ◽

Stefano Pignatti ◽

...

Keyword(s):

Machine Learning ◽

Least Squares ◽

Chlorophyll Content ◽

Learning Algorithms ◽

Machine Learning Algorithms ◽

Area Index ◽

Random Forest Tree ◽

Ground Validation ◽

Study Sites ◽

Sentinel 2

This study focuses on the comparison of hybrid methods of estimation of biophysical variables such as leaf area index (LAI), leaf chlorophyll content (LCC), fraction of absorbed photosynthetically active radiation (FAPAR), fraction of vegetation cover (FVC), and canopy chlorophyll content (CCC) from Sentinel-2 satellite data. Different machine learning algorithms were trained with simulated spectra generated by the physically-based radiative transfer model PROSAIL and subsequently applied to Sentinel-2 reflectance spectra. The algorithms were assessed against a standard operational approach, i.e., the European Space Agency (ESA) Sentinel Application Platform (SNAP) toolbox, based on neural networks. Since kernel-based algorithms have a heavy computational cost when trained with large datasets, an active learning (AL) strategy was explored to try to alleviate this issue. Validation was carried out using ground data from two study sites: one in Shunyi (China) and the other in Maccarese (Italy). In general, the performance of the algorithms was consistent for the two study sites, though a different level of accuracy was found between the two sites, possibly due to slightly different ground sampling protocols and the range and variability of the values of the biophysical variables in the two ground datasets. For LAI estimation, the best ground validation results were obtained for both sites using least squares linear regression (LSLR) and partial least squares regression, with the best performances values of R2 of 0.78, rott mean squared error (RMSE) of 0.68 m2 m−2 and a relative RMSE (RRMSE) of 19.48% obtained in the Maccarese site with LSLR. The best results for LCC were obtained using Random Forest Tree Bagger (RFTB) and Bagging Trees (BagT) with the best performances obtained in Maccarese using RFTB (R2 = 0.26, RMSE = 8.88 μg cm−2, RRMSE = 17.43%). Gaussian Process Regression (GPR) was the best algorithm for all variables only in the cross-validation phase, but not in the ground validation, where it ranked as the best only for FVC in Maccarese (R2 = 0.90, RMSE = 0.08, RRMSE = 9.86%). It was found that the AL strategy was more efficient than the random selection of samples for training the GPR algorithm.

Download Full-text