Skip to content

Latest commit

 

History

History
35 lines (22 loc) · 1.79 KB

README.md

File metadata and controls

35 lines (22 loc) · 1.79 KB

automap-soro

https://www.singularity-hub.org/static/img/hosted-singularity--hub-%23e32929.svg

This project is based on AutoMap, a pair of methods for automatic generation of evolvable genotype-phenotype mappings. Both use an artificial neural network autoencoder trained on phenotypes harvested from fitness peaks as the basis for a genotype-phenotype mapping. In the first, the decoder segment of a bottlenecked autoencoder serves as the genotype-phenotype mapping. In the second, a denoising autoencoder serves as the genotype-phenotype mapping.

The technique was introduced in

Matthew Andres Moreno, Wolfgang Banzhaf, and Charles Ofria. "Learning an Evolvable Genotype-Phenotype Mapping." Proceedings of the Genetic and Evolutionary Computation Conference. ACM, 2018.

You can find the paper and supporting materials at https://osf.io/n92c7/.

The project was built using the evosoro soft robot simulator. Evosoro was designed and developed by the Morphology, Evolution & Cognition Laboratory, University of Vermont. The library is built on top of the open source VoxCAD and the underlying voxel physics engine (Voxelyze) which were both developed by the Creative Machines Lab, Columbia University.

TODO

Experiments reported in this paper used vTODO of this software.

data, tutorials, and writeup @ https://osf.io/6jf52/

TODO

Software Authorship

Matthew Andres Moreno

[email protected]