Deep learning-based astronomical multimodal data fusion: A comprehensive review

TL;DR

This review aims to inspire and guide researchers engaged in DL-based MDF in astronomy by offering a structured overview and critical analysis and synthesizes key findings, identifies potential challenges, and suggests promising directions for future research.

Abstract

With the rapid advancements in observational technologies and the widespread implementation of large-scale sky surveys, diverse electromagnetic wave data (e.g., optical and infrared) and non-electromagnetic wave data (e.g., gravitational waves) have become increasingly accessible. Astronomy has thus entered an unprecedented era of data abundance and complexity. Astronomers have long relied on unimodal data analysis to perceive the universe, but these efforts often provide only limited insights when confronted with the current massive and heterogeneous astronomical data. In this context, multimodal data fusion (MDF), as an emerging method, provides new opportunities to enhance the value of astronomical data and deepening the understanding of the universe by integrating information from different modalities. Recent progress in artificial intelligence (AI), particularly in deep learning (DL), has greatly accelerated the development of multimodal research in astronomy. Therefore, a timely review of this field is essential. This paper begins by discussing the motivation and necessity of astronomical MDF, followed by an overview of astronomical data sources and major data modalities. It then introduces representative DL models commonly used in astronomical multimodal studies, the general fusion process as well as various fusion strategies, emphasizing their characteristics, applicability, advantages, and limitations. Subsequently, the paper surveys existing astronomical multimodal studies and datasets. Finally, the discussion section synthesizes key findings, identifies potential challenges, and suggests promising directions for future research. By offering a structured overview and critical analysis, this review aims to inspire and guide researchers engaged in DL-based MDF in astronomy.

Figures (9)

• Figure 1: The basic framework of DL-based MDF in astronomy.
• Figure 2: Multi-band and multi-platform collaborative observations. The upper part: Taking multi-band observation images of the Crab Nebula as an example, it is evident that data from different wavebands can reveal distinct details, demonstrating the complementarity among them. Credit: [$\gamma$-ray: NASA/DOE/Fermi LAT/R. Buehler; X-ray: NASA/CXC/SAO/F. Seward et al.; Ultraviolet: NASA/Swift/E. Hoversten, PSU; Optical: NASA, ESA, J. Hester, and A. Loll (Arizona State University); Infrared: NASA/JPL-Caltech/R. Gehrz (University of Minnesota); Radio: NRAO/AUI and M. Bietenholz, J. M. Uson, T. J. Cornwell]. The lower part: Given that the Earth's atmosphere absorbs most of the electromagnetic radiation, only specific wavebands (primarily including the optical band, parts of the near-infrared band, and the radio band) can effectively penetrate the atmosphere and reach the ground for detection by ground-based telescopes. For other bands such as $\gamma$-ray, X-rays, and far-infrared radiation, observations must be conducted using high-altitude balloons, aircraft, or space platforms.
• Figure 3: Data-level fusion strategy. The raw data consists of solar observation image data in the extreme ultraviolet (green) and ultraviolet (red) bands.
• Figure 4: Feature-level fusion strategy. The raw data consists of image data and spectral data of the M51 galaxy.
• Figure 5: Decision-level fusion strategy. The raw data consists of image data and spectral data of the M51 galaxy.