(149s) Parameter Estimation for Bioprocesses Cognizant of Measurement Noise Distribution

This work addresses the problem of parameter estimation for bioprocesses using Bayesian inference in a way that allows the use of measurement noise distribution to determine confidence intervals on the parameter estimates. The overarching objective is to use this information to design experiments to better estimate parameters to in-turn use for model based control. To this end, a nested sampling algorithm is utilized in conjunction with a neural network to act as a surrogate for the first principles model. The algorithm allows the use of a likelihood function that can be customized based on knowledge of the sensor noise distribution. Data from twelve runs of a bio process is used to demonstrate the proposed parameter estimation technique.

References

[1] J. Skilling, “Nested sampling for general Bayesian computation,” Bayesian Anal., vol. 1, no. 4, Dec. 2006, doi: 10.1214/06-BA127.