Skip to main content

Profile

Prof Ozgur Akman

Associate Professor of Mathematics

Email:

Telephone: 01392 724060

Extension: (Streatham) 4060

Office: Laver Building 814A

MS Bookings: Academic Personal Tutoring

GitHub: https://github.com/oeakman

Twitter: @Dr_OzgurAkman

Research interests

I am Associate Professor of Mathematics for Biology and Medicine and member of the Living Systems Mathematics group. My research focuses on the development of computational methods to systematically construct and analyse quantitative models of biochemical and neural systems. A particular focus of my work is combining machine learning with evolutionary algorithms to optimise highly-parametrised predictive models.

My doctoral work addressed the modelling of the neural networks underlying eye movement control and the tremor disorders resulting from their malfunction. Subsequently, I worked on modelling the low-level visual processing controlling attention during the viewing of multistable images. I then carried out further postdoctoral research on computational modelling of gene regulatory networks, in particular circadian clocks, during which I also worked with developmental biologists on using agent-based modelling to understand intracellular pore formation, and with computer scientists on using process algebras to represent and simulate biochemical networks.

My increasing interest in optimising high-dimensional network models led me to Exeter, where I began investigating reduced dynamic modelling to infer GRNs from multiple experimental datasets. This work was motivated by the following two questions: (i) to what extent can models of biological networks be simplified, yet still preserve predictive capacity? and (ii) how can such models be most effectively leveraged in the model construction process? A project related to this work is currently combining machine learning, single- and uni-objective evolutionary optimisation and landscape analysis to fit models of GRNs to experimental timeseries. Within this theme, I have also carried out work on using Bayesian statistical methods to efficiently infer the parameter distributions of dynamic models from experimental data. In addition, my group has been assessing the generalisability of the model-fitting algorithms developed as part of this line of research by applying them to spiking neuron models. The high-dimensional nature of the models that are central to these projects have required numerical algorithms specifically adapted to hybrid, high performance computing architectures.

Most recently, I have extended my research to encompass experimental reproducibility and the design of large-scale, automated biological experiments. The key question underpinning this work is the following: to what extent can artificial intelligence be utilised to design iterative, autonomous biotechnology experiments that are competitive with natural intelligence?

Software

Teaching

I am teaching on the following courses:

  • MTH3006: Mathematical Biology and Ecology
  • MTHM009: Advanced Topics in Mathematical and Computational Biology
  • NSCM005: Mathematical Modelling in Biology and Medicine
  • MTH2005: Modelling - Theory & Practice
  • MTHM040: MSci Mathematics Project (module leader)
  • MTHM002: Methods for Stochastics and Finance (module leader)

Recent publications