TY - JOUR
T1 - A Bayesian-inspired, deep learning-based, semi-supervised domain adaptation technique for land cover mapping
AU - Lucas, Benjamin
AU - Pelletier, Charlotte
AU - Schmidt, Daniel
AU - Webb, Geoffrey I.
AU - Petitjean, François
N1 - Publisher Copyright:
© 2021, The Author(s), under exclusive licence to Springer Science+Business Media LLC, part of Springer Nature.
PY - 2023/6
Y1 - 2023/6
N2 - Land cover maps are a vital input variable to many types of environmental research and management. While they can be produced automatically by machine learning techniques, these techniques require substantial training data to achieve high levels of accuracy, which are not always available. One technique researchers use when labelled training data are scarce is domain adaptation (DA)—where data from an alternate region, known as the source domain, are used to train a classifier and this model is adapted to map the study region, or target domain. The scenario we address in this paper is known as semi-supervised DA, where some labelled samples are available in the target domain. In this paper we present Sourcerer, a Bayesian-inspired, deep learning-based, semi-supervised DA technique for producing land cover maps from satellite image time series (SITS) data. The technique takes a convolutional neural network trained on a source domain and then trains further on the available target domain with a novel regularizer applied to the model weights. The regularizer adjusts the degree to which the model is modified to fit the target data, limiting the degree of change when the target data are few in number and increasing it as target data quantity increases. Our experiments on Sentinel-2 time series images compare Sourcerer with two state-of-the-art semi-supervised domain adaptation techniques and four baseline models. We show that on two different source-target domain pairings Sourcerer outperforms all other methods for any quantity of labelled target data available. In fact, the results on the more difficult target domain show that the starting accuracy of Sourcerer (when no labelled target data are available), 74.2%, is greater than the next-best state-of-the-art method trained on 20,000 labelled target instances.
AB - Land cover maps are a vital input variable to many types of environmental research and management. While they can be produced automatically by machine learning techniques, these techniques require substantial training data to achieve high levels of accuracy, which are not always available. One technique researchers use when labelled training data are scarce is domain adaptation (DA)—where data from an alternate region, known as the source domain, are used to train a classifier and this model is adapted to map the study region, or target domain. The scenario we address in this paper is known as semi-supervised DA, where some labelled samples are available in the target domain. In this paper we present Sourcerer, a Bayesian-inspired, deep learning-based, semi-supervised DA technique for producing land cover maps from satellite image time series (SITS) data. The technique takes a convolutional neural network trained on a source domain and then trains further on the available target domain with a novel regularizer applied to the model weights. The regularizer adjusts the degree to which the model is modified to fit the target data, limiting the degree of change when the target data are few in number and increasing it as target data quantity increases. Our experiments on Sentinel-2 time series images compare Sourcerer with two state-of-the-art semi-supervised domain adaptation techniques and four baseline models. We show that on two different source-target domain pairings Sourcerer outperforms all other methods for any quantity of labelled target data available. In fact, the results on the more difficult target domain show that the starting accuracy of Sourcerer (when no labelled target data are available), 74.2%, is greater than the next-best state-of-the-art method trained on 20,000 labelled target instances.
KW - Deep learning
KW - Domain adaptation
KW - Earth observation
KW - Land cover mapping
KW - Remote sensing
KW - Satellite image time series
UR - http://www.scopus.com/inward/record.url?scp=85102167193&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85102167193&partnerID=8YFLogxK
U2 - 10.1007/s10994-020-05942-z
DO - 10.1007/s10994-020-05942-z
M3 - Article
AN - SCOPUS:85102167193
SN - 0885-6125
VL - 112
SP - 1941
EP - 1973
JO - Machine Learning
JF - Machine Learning
IS - 6
ER -