Sanmitra Ghosh
Dr Sanmitra Ghosh
Interests
My research interests lie in the intersection of statistical inference and computational biology. Mathematical models describing biological phenomena exhibit complex behaviours such as oscillation, bifurcation, fold change to name a few. Furthermore, many biological models are known to be unidentifiable. Thus collectively, the intrinsic complexities (and also misspecifications of the modelled dynamics) present in mathematical models of biological systems render the associated inference and uncertainty quantification (UQ) problem difficult to solve. Moreover, many biological phenomena are modelled as dynamical systems described as differential equations. Repeated numerical evaluation of these systems to carry out UQ and inference tasks (especially within the Bayesian context) incurs a huge computational burden. In order to address these issues I am working on:
1) Speeding up inference methods using surrogate models.
2) Modifying current state of the art inference and UQ methods to address the intrinsic complexities of biological models.
I am also interested in concurrent developments in allied topics such as machine learning, Bayesian deep learning, probabilistic programming and their potential role in carrying out inference and UQ in computational biology.
Biography
I obtained an MSc in System on Chip from the University of Southampton in 2011, following which I pursued doctoral research in the School of Electronics and Computer Science, University of Southampton and received a PhD in Electrical & Electronics Engineering in 2016. I was supervised by Dr Srinandan Dasmahapatra and Professor Koushik Maharatna. My doctoral research focused on the application of Gaussian processes for accelerating the ABC-SMC algorithm for learning dynamical systems described as non-linear ODE/DDE. I have also used the ABC-SMC algorithm with GP acceleration to fit plant electrophysiological models to experimental data and carried out Bayesian model selection. My doctoral study is part of a EU FP7 funded project: PLEASED. I am currently a postdoctoral research assistant within the computational biology group under the supervision of Dr Gary Mirams and Professor David Gavaghan. My current research focus is on developing methods to carry out inference, UQ and experimental design for cardiac electrophysiological models.
Selected Publications
-
Fast approximate Bayesian computation for estimating parameters in differential equations
Sanmitra Ghosh‚ Srinandan Dasmahapatra and Koushik Maharatna
In Statistics and Computing. Vol. 27. No. 1. Pages 19–38. January, 2017.
Details about Fast approximate Bayesian computation for estimating parameters in differential equations | BibTeX data for Fast approximate Bayesian computation for estimating parameters in differential equations | DOI (10.1007/s11222-016-9643-4)
-
Fast approximate Bayesian computation for inference in non−linear differential equations
Sanmitra Ghosh
PhD Thesis University of Southampton. 2016.
Details about Fast approximate Bayesian computation for inference in non−linear differential equations | BibTeX data for Fast approximate Bayesian computation for inference in non−linear differential equations | Link to Fast approximate Bayesian computation for inference in non−linear differential equations
-
Drift removal in plant electrical signals via IIR filtering using wavelet energy
Saptarshi Das‚ Barry Juans Ajiwibawa‚ Shre Kumar Chatterjee‚ Sanmitra Ghosh‚ Koushik Maharatna‚ Srinandan Dasmahapatra‚ Andrea Vitaletti‚ Elisa Masi and Stefano Mancuso
In Computers and Electronics in Agriculture. Vol. 118. Pages 15 − 23. October, 2015.
Details about Drift removal in plant electrical signals via IIR filtering using wavelet energy | BibTeX data for Drift removal in plant electrical signals via IIR filtering using wavelet energy | DOI (10.1016/j.compag.2015.08.013)