Collaborative Deep Learning Models to Handle Class Imbalance in FlowCam Plankton Imagery

Kerr, T, Clark, JR, Fileman, ES, Widdicombe, CE and Pugeault, N 2020 Collaborative Deep Learning Models to Handle Class Imbalance in FlowCam Plankton Imagery. IEEE Access, 8. 170013-170032. https://doi.org/10.1109/ACCESS.2020.3022242

[img]
Preview
Text
09187202.pdf - Published Version
Available under License Creative Commons Attribution.

Download (4MB) | Preview
Official URL: http://dx.doi.org/10.1109/ACCESS.2020.3022242

Abstract/Summary

Usingautomatedimagingtechnologies,itisnowpossibletogeneratepreviouslyunprecedented volumes of plankton image data which can be used to study the composition of plankton assemblages. However, the current need to manually classify individual images introduces a bottleneck into processing chains.AlthoughMachineLearningtechniqueshavebeenusedtotryandaddressthisissue,pasteffortshave suffered from accuracy limitations, especially in minority classes. Here we use state-of-the-art methods in Deep Learning to investigate suitable architectures for training an automated plankton classification system which achieves high efficacy for both abundant and rare taxa. We collected live plankton from Station L4 in the Western English Channel and imaged 11,371 particles covering 104 taxonomic groups using the automatedplanktonimagingsystemFlowCam.Theimagesetcontainedasevereclassimbalance,withsome taxa represented by > 600 images while other, rarer taxa were represented by just 14. We demonstrate that by allowing multiple Deep Learning models to collaborate in a single classification system, classification accuracyimprovesforminorityclasseswhencomparedwiththebestindividualmodel.Thetopcollaborative model achieved a 6 % improvement in F1 accuracy over the best individual model, while overall accuracy improved by 3.2 %. This resulted in a 97.4 % overall accuracy score and a 96.2 % F1 macro score on a separate holdout test set containing 104 taxonomic groups. Based on a survey of similar studies in the literature, we believe collaborative deep learning models can significantly improve the accuracy of existing automated plankton classification systems.

Item Type: Publication - Article
Additional Keywords: Automated plankton identification, convolutional neural networks, deep learning, FlowCam, multi-layer perceptron model, Station L4.
Divisions: Plymouth Marine Laboratory > National Capability categories > Single Centre NC - CLASS
Plymouth Marine Laboratory > National Capability categories > Western Channel Observatory
Depositing User: S Hawkins
Date made live: 24 Sep 2020 14:46
Last Modified: 24 Sep 2020 14:46
URI: https://plymsea.ac.uk/id/eprint/9048

Actions (login required)

View Item View Item