Please use this identifier to cite or link to this item: https://hdl.handle.net/1959.11/59112
Title: Two-Speed Deep-Learning Ensemble for Classifcation of Incremental Land-Cover Satellite Image Patches
Contributor(s): Horry, Michael James (author); Chakraborty, Subrata  (author)orcid ; Pradhan, Biswajeet (author); Shulka, Nagesh (author); Almazroui, Mansour (author)
Publication Date: 2023-06
Open Access: Yes
DOI: 10.1007/s41748-023-00343-3
Handle Link: https://hdl.handle.net/1959.11/59112
Abstract: 

High-velocity data streams present a challenge to deep learning-based computer vision models due to the resources needed to retrain for new incremental data. This study presents a novel staggered training approach using an ensemble model comprising the following: (i) a resource-intensive high-accuracy vision transformer; and (ii) a fast training, but less accurate, low parameter-count convolutional neural network. The vision transformer provides a scalable and accurate base model. A convolutional neural network (CNN) quickly incorporates new data into the ensemble model. Incremental data are simulated by dividing the very large So2Sat LCZ42 satellite image dataset into four intervals. The CNN is trained every interval and the vision transformer trained every half interval. We call this combination of a complementary ensemble with staggered training a “two-speed” network. The novelty of this approach is in the use of a staggered training schedule that allows the ensemble model to efciently incorporate new data by retraining the high-speed CNN in advance of the resource-intensive vision transformer, thereby allowing for stable continuous improvement of the ensemble. Additionally, the ensemble models for each data increment out-perform each of the component models, with best accuracy of 65% against a holdout test partition of the RGB version of the So2Sat dataset.

Publication Type: Journal Article
Source of Publication: Earth Systems and Environment, v.7, p. 525-540
Publisher: Springer Cham
Place of Publication: Switzerland
ISSN: 2509-9434
2509-9426
Fields of Research (FoR) 2020: 4601 Applied computing
Socio-Economic Objective (SEO) 2020: tbd
Peer Reviewed: Yes
HERDC Category Description: C1 Refereed Article in a Scholarly Journal
Appears in Collections:Journal Article
School of Science and Technology

Files in This Item:
2 files
File Description SizeFormat 
openpublished/TwoSpeedChakraborty2023JournalArticle.pdfPublished Version2.48 MBAdobe PDF
Download Adobe
View/Open
Show full item record
Google Media

Google ScholarTM

Check

Altmetric


This item is licensed under a Creative Commons License Creative Commons