No project description provided
Project description
botorchex
Botorch extention library including custom acquistion functions and surrogate models.
Installation
$ pip install botorchex
Ease to use
Botorch compatible interface.
Implementation List
Custom Acqusition function
- Multi Objective Monte-Carlo Probability Improvement This acquistion function can deacrease more computational resource(wall-time) comparing to other multi objective acqusition function. This performance especially is shown in the more than 3 objctive cases. However, the convergence speed is longer than the others and there is no theoretical background.
from botorch.models.gp_regression import SingleTaskGP
from botorch.models.model_list_gp_regression import ModelListGP
from botorchex.acquisition.multi_objective.monte_carlo import qMultiProbabilityOfImprovement
model1 = SingleTaskGP(train_X, train_Y[0, :])
model2 = SingleTaskGP(train_X, train_Y[1, :])
# we assume the outputs are independent each other.
best_f = train_Y.max(dim=1)
modes = ModelListGP([model1, model2])
qPI = qMultiProbabilityOfImprovement(models, best_f)
qmpi = qMPI(test_X)
If you want to know more examples, you can check the example(multi_objective_bo.ipynb)
Custom Surrogates
- GNN based surrogates?
Referances
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
botorchex-0.1.0.tar.gz
(3.7 kB
view hashes)
Built Distribution
Close
Hashes for botorchex-0.1.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 414d44d57df921eb5c30d0b91573485ff2a69cb7b725d15d50187e43d13b5e09 |
|
MD5 | 52e9cfc5fe05e73049dec7f16f36b316 |
|
BLAKE2b-256 | 1f43366505939584aa77ca6df51036cc7971d2f67b6c911decc355a0c6cf2d95 |