AI & Machine Learning

AI and  machine learning — and deep neural networks in particular — are increasingly being used in a range of data science tasks such as data pre-processing, classification, and dimension reduction. However, these methods are far less commonly used to inform scientific inference (e.g. does x cause y? What is the functional form of this relationship? etc.). In addition to applying AI for traditional problems (e.g. machine vision), we have been exploring how tools from machine learning such as Symbolic Regression and “neuroevolution” can be used for inference in biological systems. Some applications include inferring trophic coupling in demography, inferring immigration/emigration from population time series, and studying the rules and evolution of decision-making.

Related projects:

Machine learning dynamical laws from data (w/Steve Munch and Ben Martin)

Simulated evolution using neuroevolution methods (w/Simone Olivetti)

Machine vision for underwater imagery (w/Hein Group)

Papers in this area:

Olivetti, S, MA Gil, VK Sridharan, AM HeinIn Review. Merging computational fluid dynamics and machine learning to reveal animal migration strategies.

Martin, B T, S B Munch, and A M Hein. 2018. Reverse-engineering ecological theory from data. Proc. Roy. Soc.

Hein, Rosenthal, Hagstrom, Berdahl, Torney, Couzin. 2015. The evolution of distributed sensing and collective computation in animal populations. eLife.