Ultra short period exoplanets were first discovered in 2011 using photometric data from the Kepler Space Telescope, bringing unique opportunities and challenges to the theory of planetary formation and prompting scientists to re-examine and improve existing models of planetary system formation and evolution. Ge Jian introduced that the existence of ultra short period planets provides important clues for the early evolution of planetary systems, the dynamics of planet planet interactions, and star planet interactions, including tidal forces and atmospheric erosion. For example, ultra short period planets are likely to have migrated inward from their original orbits rather than formed in their current positions - this is because the primary stars of these ultra short period planets had much larger radii in their early stages of formation than they do now, and closer ultra short period planets that were near the star during star formation are likely to have already been swallowed up by their primary stars. "At the same time, since it is often observed that ultra short period planets are accompanied by outer planets with longer period orbits, it is speculated that the origin of ultra short period planets involves the interaction between planetary brothers and sisters. These interactions relocate ultra short period planets to their current orbits close to the main star, which may be the orbits previously occupied by the stars themselves. In addition, this orbital migration of ultra short period planets may also be formed by the interaction with the protoplanetary disk or driven by the tidal interaction with the main star." However, the occurrence rate of ultra short period planets in sun like stars is very low, only about 0.5%, usually the radius is less than 2 times the Earth's radius, or in the case of super hot Jupiter, more than 10 times the Earth's radius. Radius. So far, humans have only found a total of 145 ultra short period planets, of which only 30 have radii smaller than that of Earth. Ge Jian said, "Our understanding of ultra short period planets is still very limited because the sample size is too small to accurately understand their statistical characteristics and occurrence rates." This new study provides a new method for finding ultra short period planets - the research team innovates a deep learning algorithm that combines GPU phase folding and convolutional neural networks. Professor Josh Winn, an astrophysicist at Princeton University, commented: "Ultra short period planets, also known as' lava worlds, 'have extremely extreme and unexpected properties that provide clues to our understanding of how planetary orbits change over time. I thought the transit signals in Kepler data had been exhausted and no other planets would be discovered. This technological achievement in finding new planets left a deep impression on me." Ge Jian explained that the real start time of this work was 2015. At that time, inspired by Professor Xiaolin Li from the Department of Computer Science at the University of Florida, they attempted to apply deep learning of artificial intelligence to the photometric data released by Kepler, searching for faint transit signals that could not be found by traditional methods. After nearly 10 years of effort, they finally achieved their first harvest. To use artificial intelligence to 'mine' extremely rare new discoveries in massive astronomical data, innovative artificial intelligence algorithms need to be developed. At the same time, a large number of artificial datasets generated based on the physical image features of newly discovered phenomena need to be trained to quickly, accurately, and comprehensively explore these rare and weak signals that are difficult to find in traditional ways, "said Ge Jian. (New Society)