Monday 31 March 2025
As scientists prepare for the launch of the Wide Area Vista Extragalactic Survey (WAVES) on the 4- meter Multi-Object Spectroscopic Telescope, a team of researchers has developed a innovative machine learning algorithm to ensure the survey’s success. WAVES aims to study the halo mass function at lower limits than previous surveys, and its Wide component will target galaxies with unknown redshifts.
The challenge lies in selecting the right galaxies for the survey without knowing their redshifts beforehand. Traditional methods of photometric redshift estimation are not precise enough, resulting in a completeness rate of around 90%. However, WAVES requires a much higher success rate of at least 95%.
To overcome this limitation, the researchers employed XGBoost, a tree-based classification algorithm, to predict the probability of a galaxy falling within the WAVES- Wide redshift limit. The algorithm was trained on a dataset of galaxies with known spectroscopic redshifts from overlapping surveys.
The team found that the most important features for the classification were the g-magnitude and associated g-r and u-g colors. This is likely due to the fact that the 4000 Å break passes through the VST g-band filter at the WAVES- Wide limiting redshift of 0.2.
When applied to the photometric catalog, the XGBoost algorithm achieved a purity, completeness, and F1 score of around 94.7%. The misclassified galaxies were mostly found near the redshift limit, suggesting that the algorithm is effective in identifying galaxies within the desired range.
The success of this approach has significant implications for future astronomical surveys. By combining machine learning with large datasets, scientists can improve their chances of selecting the right targets and achieving higher completeness rates. This could lead to new discoveries and a deeper understanding of the universe.
In addition to its scientific applications, the WAVES- Wide survey will also provide an opportunity to test and refine the XGBoost algorithm. As more data becomes available, the team can continue to fine-tune their approach and improve its performance.
The launch of WAVES marks an exciting new chapter in astronomical research, and the innovative use of machine learning is just one example of how scientists are pushing the boundaries of what is possible. With its unique combination of spectroscopic and photometric data, WAVES promises to reveal new insights into the structure and evolution of the universe.
Cite this article: “Unlocking the Secrets of the Universe: WAVES Survey Leverages Machine Learning”, The Science Archive, 2025.
Machine Learning, Astronomy, Spectroscopy, Photometry, Galaxy Survey, Redshift Estimation, Xgboost, Classification Algorithm, Wide Area Vista Extragalactic Survey (Waves), Completeness Rate.
