Bayesian methods for time series classification
The analysis of protein concentrations in liquids plays an important role in medical diagnostics. We combine Bayesian methods and artificial neural networks for real-time predictions of the concentrations based on observations of the corresponding chemical reactions. Our prototype model for classification of different protein concentrations reduces the prediction time by 90%. Now we are in the process of generalizing our technique for regression.
Principle Investigators: Pavel Gurevich, Hannes Stuke
Members: Julian Stastny