Investigating the Impact of the Training Set Size on Deep Learning-Powered Hyperspectral Unmixing

2021 ◽  
Author(s):  
Lukasz Tulczyjew ◽  
Jakub Nalepa
Keyword(s):  
Deep Learning ◽  
Training Set ◽  
Hyperspectral Unmixing ◽  
Set Size ◽  
The Impact

scholarly journals EVALUATION OF DEEP LEARNING TECHNIQUES FOR DEFORESTATION DETECTION IN THE AMAZON FOREST

2019 ◽  
Vol IV-2/W7 ◽  
pp. 121-128 ◽  
Author(s):  
M. X. Ortega ◽  
J. D. Bermudez ◽  
P. N. Happ ◽  
A. Gomes ◽  
R. Q. Feitosa
Keyword(s):  
Deep Learning ◽  
Support Vector ◽  
Landsat 8 ◽  
Amazon Forest ◽  
Training Set ◽  
Convolutional Network ◽  
Set Size ◽  
Learning Techniques ◽  
The Impact ◽  
Evaluation Of Methods

<p><strong>Abstract.</strong> Deforestation is one of the main causes of biodiversity reduction, climate change among other destructive phenomena. Thus, early detection of deforestation processes is of paramount importance. Motivated by this scenario, this work presents an evaluation of methods for automatic deforestation detection, specifically Early Fusion (EF) Convolutional Network, Siamese Convolutional Network (S-CNN) and the well-known Support Vector Machine (SVM), taken as the baseline. These methods were evaluated in a region of the Brazilian Legal Amazon (BLA). Two Landsat 8 images acquired in 2016 and 2017 were used in our experiments. The impact of training set size was also investigated. The Deep Learning-based approaches clearly outperformed the SVM baseline in our approaches, both in terms of F1-score and Overall Accuracy, with a superiority of S-CNN over EF.</p>

Automatic Deep Learning-Based Consolidation/Collapse Classification in Lung Ultrasound Images for COVID-19 Induced Pneumonia

2022 ◽  
Author(s):  
Nabeel Durrani ◽  
Damjan Vukovic ◽  
Maria Antico ◽  
Jeroen van der Burgt ◽  
Ruud JG van van Sloun ◽  
...  
Keyword(s):  
Deep Learning ◽  
Class Imbalance ◽  
Ultrasound Images ◽  
Pleural Effusions ◽  
Data Set ◽  
Label Noise ◽  
Set Size ◽  
Video Frames ◽  
Two Factors ◽  
The Impact

<div>Our automated deep learning-based approach identifies consolidation/collapse in LUS images to aid in the diagnosis of late stages of COVID-19 induced pneumonia, where consolidation/collapse is one of the possible associated pathologies. A common challenge in training such models is that annotating each frame of an ultrasound video requires high labelling effort. This effort in practice becomes prohibitive for large ultrasound datasets. To understand the impact of various degrees of labelling precision, we compare labelling strategies to train fully supervised models (frame-based method, higher labelling effort) and inaccurately supervised models (video-based methods, lower labelling effort), both of which yield binary predictions for LUS videos on a frame-by-frame level. We moreover introduce a novel sampled quaternary method which randomly samples only 10% of the LUS video frames and subsequently assigns (ordinal) categorical labels to all frames in the video based on the fraction of positively annotated samples. This method outperformed the inaccurately supervised video-based method of our previous work on pleural effusions. More surprisingly, this method outperformed the supervised frame-based approach with respect to metrics such as precision-recall area under curve (PR-AUC) and F1 score that are suitable for the class imbalance scenario of our dataset despite being a form of inaccurate learning. This may be due to the combination of a significantly smaller data set size compared to our previous work and the higher complexity of consolidation/collapse compared to pleural effusion, two factors which contribute to label noise and overfitting; specifically, we argue that our video-based method is more robust with respect to label noise and mitigates overfitting in a manner similar to label smoothing. Using clinical expert feedback, separate criteria were developed to exclude data from the training and test sets respectively for our ten-fold cross validation results, which resulted in a PR-AUC score of 73% and an accuracy of 89%. While the efficacy of our classifier using the sampled quaternary method must be verified on a larger consolidation/collapse dataset, when considering the complexity of the pathology, our proposed classifier using the sampled quaternary video-based method is clinically comparable with trained experts and improves over the video-based method of our previous work on pleural effusions.</div>

Deep Learning Stream Temperature Model: Recommendations for Modeling Gauged and Ungauged Basins

2021 ◽  
Author(s):  
Farshid Rahmani ◽  
Kathryn Lawson ◽  
Samantha Oliver ◽  
Alison Appling ◽  
Chaopeng Shen
Keyword(s):  
Deep Learning ◽  
Root Mean Square Error ◽  
Mean Square Error ◽  
Root Mean Square ◽  
The United States ◽  
Data Availability ◽  
Efficiency Coefficient ◽  
Mean Square ◽  
Training Set ◽  
The Impact

&lt;p&gt;Stream water temperature (T&lt;sub&gt;s&lt;/sub&gt;) is a variable that plays a pivotal role in managing water resources. We used the long short-term memory (LSTM) deep learning architecture to develop a basin centric single T&lt;sub&gt;s&lt;/sub&gt; model based on general meteorological data and basin meteo-geological attributes. We created a strong tool for long-term Ts projection and subsequently, improved the Ts model using novel approaches. We investigated the impact of both observed and simulated streamflow data on improving the model accuracy. At a national scale, we obtained a median root-mean-square error (RMSE) of 0.69 &lt;sup&gt;o&lt;/sup&gt;C, and Nash-Sutcliffe model efficiency coefficient (NSE) of 0.985, which are marked improvements over previous values reported in previous studies. In order to test the performance of the model on basins ranging from basins with extensive data to unmonitored basins, we used more than 400 basins with different data-availability groups (DAG) across the continent of the United States to explore how to assemble the training dataset for both monitored and unmonitored basins. Best root-mean-square error (RMSE) for sites with extensive (99%), intermediate (60%), scarce (10%) and absent (0%) data for training were 0.75, 0.837, 0.889, and 1.595 &lt;sup&gt;o&lt;/sup&gt;C, respectively. We observed the negative effect of the presence of reservoirs in T&lt;sub&gt;s&lt;/sub&gt; modeling. Our results illustrated that the most suitable training set should be different in modeling basins with different availability of observed data. for predicting T&lt;sub&gt;s&lt;/sub&gt; in a monitored basin, including basins that have at least equal DAG with that particular basin will result in most accurate predictions, however, for T&lt;sub&gt;s&lt;/sub&gt; prediction in ungauged basin, including all basins in training section will generate the best model, showing a more diverse training set. Furthermore, to decrease overfitting produced by attributes for PUB application, we could improve the accuracy of the model using input-selection ensemble method. We got median correlation higher than 0.90 for PUB after seasonality was removed which is still high. While many T&lt;sub&gt;s&lt;/sub&gt; prediction models showed better performance in summer, our model was on the opposite side. We found a strong relationship between general available daily meteorological variables and catchment attributes with the presented T&lt;sub&gt;s&lt;/sub&gt; model. However, our results indicate that combining physics-based criteria to the model can improve the prediction of temperature in river networks.&lt;/p&gt;&lt;p&gt;.&lt;/p&gt;

scholarly journals Tens of images can suffice to train neural networks for malignant leukocyte detection

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Jens P. E. Schouten ◽  
Christian Matek ◽  
Luuk F. P. Jacobs ◽  
Michèle C. Buck ◽  
Dragan Bošnački ◽  
...  
Keyword(s):  
Neural Networks ◽  
Deep Learning ◽  
Single Cell ◽  
Expert Knowledge ◽  
Binary Classification ◽  
Lymphoblastic Leukemia ◽  
Training Set ◽  
Early Detection Of Disease ◽  
Set Size ◽  
Training Images

AbstractConvolutional neural networks (CNNs) excel as powerful tools for biomedical image classification. It is commonly assumed that training CNNs requires large amounts of annotated data. This is a bottleneck in many medical applications where annotation relies on expert knowledge. Here, we analyze the binary classification performance of a CNN on two independent cytomorphology datasets as a function of training set size. Specifically, we train a sequential model to discriminate non-malignant leukocytes from blast cells, whose appearance in the peripheral blood is a hallmark of leukemia. We systematically vary training set size, finding that tens of training images suffice for a binary classification with an ROC-AUC over 90%. Saliency maps and layer-wise relevance propagation visualizations suggest that the network learns to increasingly focus on nuclear structures of leukocytes as the number of training images is increased. A low dimensional tSNE representation reveals that while the two classes are separated already for a few training images, the distinction between the classes becomes clearer when more training images are used. To evaluate the performance in a multi-class problem, we annotated single-cell images from a acute lymphoblastic leukemia dataset into six different hematopoietic classes. Multi-class prediction suggests that also here few single-cell images suffice if differences between morphological classes are large enough. The incorporation of deep learning algorithms into clinical practice has the potential to reduce variability and cost, democratize usage of expertise, and allow for early detection of disease onset and relapse. Our approach evaluates the performance of a deep learning based cytology classifier with respect to size and complexity of the training data and the classification task.

Automatic Deep Learning-Based Consolidation/Collapse Classification in Lung Ultrasound Images for COVID-19 Induced Pneumonia

2022 ◽  
Author(s):  
Nabeel Durrani ◽  
Damjan Vukovic ◽  
Maria Antico ◽  
Jeroen van der Burgt ◽  
Ruud JG van van Sloun ◽  
...  
Keyword(s):  
Deep Learning ◽  
Class Imbalance ◽  
Ultrasound Images ◽  
Pleural Effusions ◽  
Data Set ◽  
Label Noise ◽  
Set Size ◽  
Video Frames ◽  
Two Factors ◽  
The Impact

<div>Our automated deep learning-based approach identifies consolidation/collapse in LUS images to aid in the diagnosis of late stages of COVID-19 induced pneumonia, where consolidation/collapse is one of the possible associated pathologies. A common challenge in training such models is that annotating each frame of an ultrasound video requires high labelling effort. This effort in practice becomes prohibitive for large ultrasound datasets. To understand the impact of various degrees of labelling precision, we compare labelling strategies to train fully supervised models (frame-based method, higher labelling effort) and inaccurately supervised models (video-based methods, lower labelling effort), both of which yield binary predictions for LUS videos on a frame-by-frame level. We moreover introduce a novel sampled quaternary method which randomly samples only 10% of the LUS video frames and subsequently assigns (ordinal) categorical labels to all frames in the video based on the fraction of positively annotated samples. This method outperformed the inaccurately supervised video-based method of our previous work on pleural effusions. More surprisingly, this method outperformed the supervised frame-based approach with respect to metrics such as precision-recall area under curve (PR-AUC) and F1 score that are suitable for the class imbalance scenario of our dataset despite being a form of inaccurate learning. This may be due to the combination of a significantly smaller data set size compared to our previous work and the higher complexity of consolidation/collapse compared to pleural effusion, two factors which contribute to label noise and overfitting; specifically, we argue that our video-based method is more robust with respect to label noise and mitigates overfitting in a manner similar to label smoothing. Using clinical expert feedback, separate criteria were developed to exclude data from the training and test sets respectively for our ten-fold cross validation results, which resulted in a PR-AUC score of 73% and an accuracy of 89%. While the efficacy of our classifier using the sampled quaternary method must be verified on a larger consolidation/collapse dataset, when considering the complexity of the pathology, our proposed classifier using the sampled quaternary video-based method is clinically comparable with trained experts and improves over the video-based method of our previous work on pleural effusions.</div>

Classical scoring functions for docking are unable to exploit large volumes of structural and interaction data

2019 ◽  
Vol 35 (20) ◽  
pp. 3989-3995 ◽  
Author(s):  
Hongjian Li ◽  
Jiangjun Peng ◽  
Pavel Sidorov ◽  
Yee Leung ◽  
Kwong-Sak Leung ◽  
...  
Keyword(s):  
Machine Learning ◽  
Protein Structures ◽  
Superior Performance ◽  
Supplementary Information ◽  
Scoring Functions ◽  
Training Set ◽  
Test Set ◽  
Set Size ◽  
Extreme Gradient Boosting ◽  
The Impact

Abstract Motivation Studies have shown that the accuracy of random forest (RF)-based scoring functions (SFs), such as RF-Score-v3, increases with more training samples, whereas that of classical SFs, such as X-Score, does not. Nevertheless, the impact of the similarity between training and test samples on this matter has not been studied in a systematic manner. It is therefore unclear how these SFs would perform when only trained on protein-ligand complexes that are highly dissimilar or highly similar to the test set. It is also unclear whether SFs based on machine learning algorithms other than RF can also improve accuracy with increasing training set size and to what extent they learn from dissimilar or similar training complexes. Results We present a systematic study to investigate how the accuracy of classical and machine-learning SFs varies with protein-ligand complex similarities between training and test sets. We considered three types of similarity metrics, based on the comparison of either protein structures, protein sequences or ligand structures. Regardless of the similarity metric, we found that incorporating a larger proportion of similar complexes to the training set did not make classical SFs more accurate. In contrast, RF-Score-v3 was able to outperform X-Score even when trained on just 32% of the most dissimilar complexes, showing that its superior performance owes considerably to learning from dissimilar training complexes to those in the test set. In addition, we generated the first SF employing Extreme Gradient Boosting (XGBoost), XGB-Score, and observed that it also improves with training set size while outperforming the rest of SFs. Given the continuous growth of training datasets, the development of machine-learning SFs has become very appealing. Availability and implementation https://github.com/HongjianLi/MLSF Supplementary information Supplementary data are available at Bioinformatics online.

Removal of multisource noise in airborne electromagnetic data based on deep learning

Geophysics ◽  
2020 ◽  
Vol 85 (6) ◽  
pp. B207-B222
Author(s):  
Xin Wu ◽  
Guoqiang Xue ◽  
Yiming He ◽  
Junjie Xue
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Noise Removal ◽  
Specific Method ◽  
Noise Power ◽  
Quality Of Data ◽  
Training Set ◽  
Useful Signal ◽  
Airborne Electromagnetic ◽  
The Impact

Existing noise removal processes for airborne electromagnetic (AEM) data generally consist of several steps, with each using a specific method to remove a specific type of noise. To improve the efficiency of AEM denoising and reduce the impact of the subjective judgment of the operators on the processing results, we have adopted a deep learning method based on a denoising autoencoder (DAE), which enables in one single processing step the removal of multisource noise. The most common noise sources in AEM data, including motion-induced noise, nearby or moderately distant sferics noise, power-line noise, and background electromagnetic noise, will be combined with a large number of simulation responses to build a training set. The data in the training set will be used to train the deep learning DAE neural network so that the neural network could fully learn the respective characteristics of the signal and noise and further effectively distinguish the AEM response signal (useful signal) from the above noise. The field data were processed using this method, and the processing results were compared with those obtained using traditional methods. The comparison test revealed that this method is helpful to reduce the influence of subjective factors on the quality of data results and compress the entire AEM data processing time.

scholarly journals Deep‐Learning‐Based Neural Tissue Segmentation of MRI in Multiple Sclerosis: Effect of Training Set Size

2019 ◽  
Vol 51 (5) ◽  
pp. 1487-1496 ◽  
Author(s):  
Ponnada A. Narayana ◽  
Ivan Coronado ◽  
Sheeba J. Sujit ◽  
Jerry S. Wolinsky ◽  
Fred D. Lublin ◽  
...  
Keyword(s):  
Multiple Sclerosis ◽  
Deep Learning ◽  
Neural Tissue ◽  
Tissue Segmentation ◽  
Training Set ◽  
Set Size

scholarly journals The effect of training set size in authorship attribution: application on short arabic texts

2019 ◽  
Vol 9 (1) ◽  
pp. 652
Author(s):  
Mohammed Al-Sarem ◽  
Abdel-Hamid Emara
Keyword(s):  
Linear Regression ◽  
Mahalanobis Distance ◽  
Authorship Attribution ◽  
Training Dataset ◽  
Training Set ◽  
Research Papers ◽  
Set Size ◽  
Dataset Size ◽  
The Impact ◽  
And Training

<span lang="EN-US">Authorship attribution (AA) is a subfield of linguistics analysis, aiming to identify the original author among a set of candidate authors. Several research papers were published and several methods and models were developed for many languages. However, the number of related works for Arabic is limited. Moreover, investigating the impact of short words length and training set size is not well addressed. To the best of our knowledge, no published works or researches, in this direction or even in other languages, are available. Therefore, we propose to investigate this effect, taking into account different stylomatric combination. The Mahalanobis distance (MD), Linear Regression (LR), and Multilayer Perceptron (MP) are selected as AA classifiers. During the experiment, the training dataset size is increased and the accuracy of the classifiers is recorded. The results are quite interesting and show different classifiers behaviours. Combining word-based stylomatric features with n-grams provides the best accuracy reached in average 93%.</span>

Sign in / Sign up

Export Citation Format

Share Document