Exploring the potential of transfer learning for metamodels of heterogeneous material deformation

Explore this paper's citation graph

Summary

This paper extends Mechanical MNIST, an open source benchmark dataset of heterogeneous material undergoing large deformation, to include a selection of low-fidelity simulation results that require ≈ 2 - 4 orders of magnitude less CPU time to run, and shows that transfer learning can vastly improve the performance of metamodels used to predict the results of high-f fidelity simulations.

Type
preprint
Published
2020-10-28
Cited by
20
References
63
Access
Open access

Keywords

Computer science, Leverage (statistics), Fidelity, Transfer of learning, MNIST database

References

Cited by

Related papers