The Miracle of Artificial Intelligence: Chinese Scientists Extend Vision to Galaxies 13 Billion Light-Years Away

Beijing — Chinese researchers have achieved a historic leap in astronomy by developing an artificial intelligence (AI) model capable of reading subtle light signals hidden in the deepest, darkest corners of the universe. An interdisciplinary research team from Tsinghua University has created a model named ‘ASTERIS’ (Astronomical Spatiotemporal Enhancement and Reconstruction for Image Synthesis), combining computational optics and advanced AI algorithms.

According to a study published in the internationally renowned journal Science on Friday, this model can extract extremely faint astronomical signals, identify galaxies more than 13 billion light-years away, and produce the deepest space images ever captured.

Studying distant and extremely dim celestial objects is considered essential for understanding the origin and evolution of the universe. However, astronomers have long faced a significant challenge: the weak light signals from distant objects are often obscured by background sky noise and thermal radiation emitted by telescopes themselves.

To address this challenge, the ASTERIS model employs a “self-supervised spatiotemporal denoising” technique. When the research team applied this technique to data from the James Webb Space Telescope (JWST), they discovered that the observable range expanded from visible light at 500 nanometers to the mid-infrared region up to 5 micrometers. This has increased the observation depth by 1.0 magnitude, meaning objects 2.5 times fainter than before can now be detected.

Using ASTERIS, the researchers have identified over 160 potential high-redshift galaxies formed during the “Cosmic Dawn”—the period approximately 200 million to 500 million years after the Big Bang. This number is three times higher than detections made using previous methods.

According to team member and Associate Professor Zheng Cai from the Department of Astronomy at Tsinghua University, “This technology has opened a new door for studying the early universe. We can now clearly see signals we previously thought impossible to detect.”

Traditional noise reduction techniques rely on ‘stacking’ multiple images together, based on the assumption that the noise is similar. However, noise in deep space varies over time and space. ASTERIS solves this problem by reconstructing deep space images as three-dimensional space-time volumes.

The “photometric adaptive screening mechanism” used in this model distinguishes subtle noise fluctuations, enabling the identification of ultra-faint signals from distant stars and galaxies.

A reviewer of the research commented that this is a “highly relevant work that could have a profound impact on various fields of astronomy.” Similarly, Professor Dai Qionghai from the Department of Automation at Tsinghua University stated, “It is now possible to reconstruct faint celestial objects obscured by optical noise with high fidelity.”

The researchers stated that this AI model is capable of analyzing massive amounts of space telescope data and is adaptable to various observation platforms. Therefore, it holds the potential to be developed as a universal deep-space data enhancement platform in the future.

It is anticipated that applying this technology in next-generation telescopes will help answer major scientific questions, unraveling the mysteries of dark energy, dark matter, the origin of the universe, and exoplanets.

In summary, ASTERIS has not only enhanced images; it has extended humanity’s vision to 13 billion light-years away, providing an opportunity to read the primordial story of the universe with greater clarity.