When optimizing ML model, sub-models are generated at each loop. They all also project features of the training set. From them, the model's probability distribution was acquired and the uncertainty of the model prediction was evaluated.
Wang, Z., et al. (2010). "Large variations in Southern Hemisphere biomass burning during the last 650 years." Science 330(6011): 1663-1666. Biomass burning and methane oxidation affect the amount of CO. C derived from biomass burning has higher delta C-13 than methane oxidized C. Also, there is fraction between O-16 and O-18. Using delta C-13 and delta O-18, CO amounts and their relative source can be retrieved.
댓글
댓글 쓰기