Machine learning predicts copper nanoparticle synthesis with small datasets

Researchers have developed a machine learning approach to predict the hydrodynamic diameter of copper nanoparticles (Cu NPs) using a small dataset of 25 syntheses. This method aims to overcome the challenges of reproducible and size-controlled synthesis, which are often hindered by the scarcity of large experimental data sets. The study utilized Latin Hypercube Sampling for efficient parameter space coverage and employed ensemble regression models that demonstrated comparable generalization to traditional Design of Experiments (DoE) models, achieving a higher R2 score of 0.74 compared to the DoE model's 0.60. While classification models using random forests and LLMs were also evaluated, their performance was modest, suggesting that complex models like LLMs may not offer significant benefits with such limited data. AI

IMPACT Demonstrates the potential of carefully curated small datasets and classical ML for guiding synthesis in materials science, suggesting LLMs may not be optimal for such constrained data scenarios.

RANK_REASON The cluster is a research paper detailing a new methodology for materials science using machine learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Danny E. P. Vanpoucke
• DLS
• Ensemble regression models
• Latin Hypercube Sampling
• LLMs
• machine learning
• random forest
• United States Department of Energy

• paper
• model release

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