IEEE Geoscience and Remote Sensing Magazine - June 2021 - 7

FROM THE GUEST EDITORS
BY SABAH MOHAMMED, TAI-HOON KIM,
PEDRAM GHAMISI, AND RUAY-SHIUNG CHANG
A Special Issue on Recent Progress in
Developing Artificial Intelligence and
Machine Learning Methodologies
A
rtificial intelligence (AI) plays a growing role in geoscience
and remote sensing with thousands-of-terabyte
images collected every day from satellites, radars,
aerial vehicles, drones, aircrafts, and other sensing devices.
Machine learning (ML) and other AI technologies
are challenged to use these data sets and predict patterns
related to atmospheric, environmental, oceanographic,
and other changes to the land and the ecosystem of our
planet. In the past, remote sensing proved to be successful
with typical applications like estimating the typhoon
rainfall over oceans, monitoring reservoir water quality,
mapping base-metal deposits, landslide detection,
soil moisture distribution, crop-type classification, vegetation
change detection, and analyzing geological
structures. However, modern ML and AI have shifted
the focus of remote sensing from using mathematical
and statistical techniques for image processing to
integrating sophisticated computational intelligence
and embedding higher knowledge. The use of neural
networks and the power of deep learning have become
popular in geoscience over the recent past [1], providing
greater insights for interesting discovery of geosciences
patterns. Although neural networks were criticized earlier
by some geoscientists as a black box engine where
the decision-making process is uninterpretable, recent
neurosymbolic AI has provided the solution for this obstacle
[2]. Moreover, the greater involvement of expert
tagging and annotation of the collected data through
crowdsourcing [3], integrative workspaces [4], and thick
data analytics [5] has broadened the availability of good
training data sets for solving diverse geoscience problems.
Figure 1 illustrates some of the new trends in using
AI and ML in geosciences and remote sensing.
Digital Object Identifier 10.1109/MGRS.2021.3078373
Date of current version: 17 June 2021
JUNE 2021 IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE
This special issue focuses on recent progress in developing
AI and ML methodologies with a variety of Earth
science and remote sensing applications. The modern
geoscientist needs the power of AI and ML to conduct
explorations and predict Earth changes through the use
of large numbers of data sets and expertise in smart and
intelligent systems. This special issue received nine submissions,
from which we have accepted four.
The first article, by Zhou et al., gives a comprehensive
overview of AI-based phase unwrapping (PU) techniques
in interferometric synthetic aperture radar (InSAR).
First, the authors survey AI-based, single-baseline
PU methods; then, they review AI techniques related to
multibaseline PU. In addition, they show several experimental
examples of these, methods from both simulated
and real InSAR data sets, providing readers with
an overview of AI-based PU processing's potential
and limits.
In the second article, Fracastoro et al. introduce a survey
of the literature on deep learning methods applied
to SAR despeckling, covering both supervised and more
recent self-supervised approaches. The authors provide
a critical analysis of existing methods with the objective
of recognizing the most promising research lines,
identifying the factors that have limited the success
of deep models, and proposing ways forward in an attempt
to fully exploit the potential of deep learning for
SAR despeckling.
In third article, Hong et al. present an advanced and
cutting-edge technical survey for nonconvex modeling
toward interpretable AI models, covering a broad scope
on the topic of hyperspectral (HS) remote sensing, which
includes HS image restoration, dimensionality reduction,
data fusion and enhancement, spectral unmixing,
and cross-modality learning for large-scale land cover
mapping. Around these topics, the article showcases the
significance of nonconvex techniques to bridge the gap
7
IEEE Geoscience and Remote Sensing Magazine - June 2021

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2021
