부산대학교 도서관

This study digs into the ever-changing environment of Neural Architecture Search (NAS) and Atom (Automated Machine Learning) for the purpose of deep learning model optimization. We take a methodical look at these methods to see whether they can really bring about a sea change in the way AI is created and used. Our studies are based on a comprehensive process that covers issue statements, data cleaning, Atom and NAS algorithm selection, optimization goals, model training, and ethical concerns. We offer simulated experimental findings that demonstrate the effectiveness of these techniques, focusing on the dramatic improvements to model performance and the time savings from automation. Importantly, our research highlights the need for justice, accountability, and openness in the context of automated model optimization. Our results have important implications for the field of artificial intelligence and for society at general, which we explain as part of our conclusion. We conclude by outlining potential avenues for further study, such as the use of transfer learning, scalability, hybrid techniques, and self-adaptive algorithms. A thorough introduction to Atom and NAS, this study provides valuable ideas that may promote growth, accessibility, and ethical responsibility in AI.