Machine learning for expert‐level image‐based identification of very similar species in the hyperdiverse plant bug family Miridae (Hemiptera: Heteroptera)

There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

Abstract

Deep learning algorithms and particularly convolutional neural networks are very successful in pattern recognition from images and are increasingly employed in biology. The development of automated systems for rapid and reliable species identification is vital for insect systematics and may revolutionize this field soon. In this study, we demonstrate the ability of a convolutional neural network to identify species based on habitus photographs with expert‐level accuracy in a taxonomically challenging group where a human‐based identification would require notorious genitalia dissections. Using the economically important and polymorphic plant bug genus Adelphocoris Reuter (Heteroptera: Miridae) as a model group, we explore the variability in the performance of 11 convolutional neural models most commonly used for image classification, test the role of class‐imbalance on the model performance assessment and visualize areas of interest using three interpretation algorithms. Classification performance in our experiments with collection‐based habitus photographs is high enough to identify very similar species from a large group with an expert‐level accuracy. The accuracy is getting lower only in the experiments with an additional dataset of Adelphocoris and other live plant bugs photographs taken from the Web. Our article demonstrates the importance of comprehensive institutional insect collections for bringing deep learning algorithms into service for systematic entomology using affordable equipment and methods.

Related collections

Most cited references 62

Record: found
Abstract: not found
Conference Proceedings: not found

Deep Residual Learning for Image Recognition

Kaiming He, Xiangyu Zhang, Shaoqing Ren … (2019)

0 comments Cited 8212 times – based on 0 reviews

Bookmark

Record: found
Abstract: found
Article: not found

SMOTE: Synthetic Minority Over-sampling Technique

N. Chawla, K. W. Bowyer, L Hall … (2002)

An approach to the construction of classifiers from imbalanced datasets is described. A dataset is imbalanced if the classification categories are not approximately equally represented. Often real-world data sets are predominately composed of ``normal'' examples with only a small percentage of ``abnormal'' or ``interesting'' examples. It is also the case that the cost of misclassifying an abnormal (interesting) example as a normal example is often much higher than the cost of the reverse error. Under-sampling of the majority (normal) class has been proposed as a good means of increasing the sensitivity of a classifier to the minority class. This paper shows that a combination of our method of over-sampling the minority (abnormal) class and under-sampling the majority (normal) class can achieve better classifier performance (in ROC space) than only under-sampling the majority class. This paper also shows that a combination of our method of over-sampling the minority class and under-sampling the majority class can achieve better classifier performance (in ROC space) than varying the loss ratios in Ripper or class priors in Naive Bayes. Our method of over-sampling the minority class involves creating synthetic minority class examples. Experiments are performed using C4.5, Ripper and a Naive Bayes classifier. The method is evaluated using the area under the Receiver Operating Characteristic curve (AUC) and the ROC convex hull strategy.