The recent focus in news has been on the progress of artificial intelligence (AI) in the past couple of years. ChatGPT and DALL·E are examples of AI models that many people associate with AI. AI tools are utilized by astronomers to analyze vast data sets, which would be impractical to manually go through. Machine Learning Algorithms (ML) are crucial for categorizing data based on predetermined parameters derived from previous studies. An example of ML usage is in the identification of elusive patterns in sky surveys by astronomers, though the limitations of this method in classifying objects in space are not thoroughly understood.
To address these limitations, a group of scientists led by Pamela Marchand-Cortes at the University of La Serena in Chile tested the capabilities of ML. They used ML models like Rotation forest, Random forest, and Logit Boost to categorize objects beyond the Milky Way galaxy based on their properties. The team aimed to see if ML could accurately categorize objects already manually classified. The challenge was in the dense region of sky obscured by dust in the Milky Way, known as the “Avoidance Zone.” The team’s experiment showed that ML had difficulty in categorizing objects in this challenging area.
The team gathered and analyzed data from X-ray images to manually identify objects and compare ML’s performance. ML correctly identified large objects like galaxies in only a few instances, showcasing its limitations. Despite the potential for ML to assist in studying obscured regions of the universe, the team recommended training AI models with diverse samples to enhance accuracy in future research.
Post View: 120
Source: sciworthy.com
