Automated deep learning

Deep learning models usually come with a large number of hyperparameters, and their selection is critical for their performance. To find the set of hyperparameters that generates optimal model performance, one needs to employ hyperparameter tuning, or optimisation (HPO).

This project developed HPO with Bayesian optimisation for the automated hyperparameter tunning for convolutional neural networks (CNN) (Shen & Viscarra Rossel, 2021).

Building blocks of CNN. The number of different blocks and their internal hyperparameters are optimised.

References

2021

Sci. Rep.

Automated spectroscopic modelling with optimised convolutional neural networks

Zefang Shen, and RA Viscarra Rossel

Scientific Reports, 2021

Bib HTML

@article{shen2021automated,
  title = {Automated spectroscopic modelling with optimised convolutional neural networks},
  author = {Shen, Zefang and Viscarra Rossel, RA},
  journal = {Scientific Reports},
  volume = {11},
  number = {1},
  pages = {208},
  year = {2021},
  publisher = {Nature Publishing Group UK London},
}