Research opportunities exist for PhD students interested in AI/ML for computational science; more specifically, in “Mathematical AI/ML for Multiscale Computational Science” the latter including computational biology. Please note that this is quite a specialized niche, not to be confused with general machine learning, AI, or bioinformatics. It also requires quite a bit of background knowledge across several domains. On the other hand it could turn out to be the future of science.
Prospective PhD students should apply to one of these graduate programs:
– Computer Science
– Mathematical, Computational, and Systems Biology,
specifically mentioning my name in your application. (There is also a forthcoming joint degree program in Computational Science, not yet up and running as of 11/2017.) Please note that my current research approaches all require a strong mathematical background and an interest in computational science. Here are some of my main current project interest areas …
Possible Research Topics for Prospective PhD Students,
in “Mathematical AI/ML for Multiscale Computational Science”:
1. Machine learning for change-of-scale relationships in predictive dynamical models arising in mathematical computational biology and computational science .
2. Rewrite rules for labeled graphs and symbolic structures, applied to e.g. (a) dynamical model formulation ; (b) extended-object type construction for dynamical models ; (c) symbolic mathematical analysis of dynamical models by proof assistant software based on type theory; and/or (d) model transformation by semantics-preserving or semantics-approximating chainable meta-rules (e.g. those of the ML project #1 above). This project will bring mathematical forms of symbolic AI, including computer algebra, to bear on complex scientific modeling.
3. High-level mathematical/scientific modeling languages and systems targeting high-performance computing (HPC) for predictive dynamical models, in computational biology and computational science. This project could incorporate the results of projects 1 and/or 2 above.
4. Biomorphic neural network growth and evolution, using recent brain architecture knowledge from model organisms. This project could, but need not, incorporate the results of projects 1-3 above.
Other projects may also be possible; see my recent publications for current interests.
Caution: Funding for these projects may take the form of teaching assistantships and/or student-obtained funding (e.g. student-obtained fellowships), unless future grants come through.
Postdoctoral students interested in the foregoing specialized topics and qualified in the relevant fields named above would currently have to bring their own funding, e.g. through fellowships. Having such funding, however, does not guarantee acceptance as a postdoc.
 Johnson et. al. 2015 “Model Reduction for Stochastic CaMKII Reaction Kinetics in Synapses by Graph-Constrained Correlation Dynamics”, Todd Johnson, Thomas Bartol, Terrence Sejnowski, and Eric Mjolsness. Physical Biology 12:4, July 2015.
 Mjolsness and Yosiphon 2006 Stochastic Process Semantics for Dynamical Grammars. Eric Mjolsness and Guy Yosiphon, Annals of Mathematics and Artificial Intelligence, 47(3-4) August 2006.
 Mjolsness 2010 “Towards Measurable Types for Dynamical Process Modeling Languages”, Eric Mjolsness. Electronic Notes in Theoretical Computer Science (ENTCS), vol. 265, pp. 123-144, 6 Sept. 2010, Elsevier. Also Proceedings of the 26th Conference on Mathematical Foundations of Programming Semantics (MFPS 2010). DOI 10.1016/j.entcs.2010.08.008.