Skip to content

Siamese Neural Networks for Regression: Similarity-Based Pairing and Uncertainty Quantification

License

Notifications You must be signed in to change notification settings

AstraZeneca/Siamese-Regression-Pairing

Repository files navigation

Similarity based pairing for Siamese Neural Network

Maturity level-0

This repository corresponds to the article'Siamese Neural Networks for Regression: Similarity-Based Pairing and Uncertainty Quantification'. The project consists of 4 models: Multilayer perceptron model with single input (MLP-FP) and paired input (MLP-deltaFP), Chemformer model[1] , and Siamese model with Chemformer strcuture (Cheformer-snn)

MLP-deltaFP:

  • for exaustive pairs:

python mlp.py -s lipo_all.yml -st 0 -f lipo

  • for similarity-based pairs:

python mlp.py -s lipo_top1.yml -st 1 -f lipo

MLP-FP:

  • python mlp.py -s lipo_mlp.yml

Cheformer-snn:

  • for dropout 0.0:

python finetuenRegr_k_fold.py --name lipo --data_path lipo/ --drp 0.0

we need to run dropout = [0.0,0.05,0.1,0.17]

Chemformer:

  • python finetuneRegr_k_fold.py --name lipo --data_path lipo/

generate plots:

  • python confidence_plot.py

  • python dropout_plot.py

  • python plot_n_shot.py

  • python shot_plot.py

[1] Irwin, R., Dimitriadis, S., He, J., Bjerrum, E.J., 2021. Chemformer: A Pre-Trained Transformer for Computational Chemistry. Mach. Learn. Sci. Technol. https://doi.org/10.1088/2632-2153/ac3ffb

About

Siamese Neural Networks for Regression: Similarity-Based Pairing and Uncertainty Quantification

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published