AdamNX algorithm enhances ML training with novel decay mechanism · 1 source tracked

Researchers have introduced AdamNX, an optimization algorithm designed to improve upon the existing Adam method for training machine learning models. This new algorithm incorporates a novel exponential decay mechanism for the second-order moment estimate, which aims to make training updates more closely resemble momentum-SGD during plateau phases. The paper presents results across image classification, object detection, and semantic segmentation tasks, noting that these findings do not establish broad statistical effects or generalization beyond the reported configurations. The code for AdamNX has been made open-source. AI

IMPACT Introduces a novel optimization technique that could improve training efficiency and performance for various machine learning tasks.

RANK_REASON The item is a research paper detailing a new algorithm for machine learning optimization. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• Adam
• AdamNX
• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• Meng Zhu
• momentum-SGD
• ScienceCast

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →