Machine Learning vs. Databases: A question of speed, accuracy and scalability
Today, the conventional method for analysing spectroscopic microplastics data is spectral library search. However, the application of spectral libraries for microplastics analysis creates many difficulties, such as setting thresholds for HQI indices or building a suitable reference database. The large set of possible parameter settings hampers reproducibility and comparability.
At the Agilent Microplastics in the Environment Virtual Symposium 2021 Co-Founder Benedikt Hufnagl will give a presentation which aims at broadening the view on alternatives based on model-based machine learning and how these methods can by-pass these problems by providing parameterless, broadly applicable analysis solutions for spectroscopy.