Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/aspuru-guzik-group/gemini
scalable multi-fidelity machine learning
https://github.com/aspuru-guzik-group/gemini
Last synced: 22 days ago
JSON representation
scalable multi-fidelity machine learning
- Host: GitHub
- URL: https://github.com/aspuru-guzik-group/gemini
- Owner: aspuru-guzik-group
- License: mit
- Created: 2021-04-03T13:11:24.000Z (about 3 years ago)
- Default Branch: main
- Last Pushed: 2021-07-16T19:30:20.000Z (almost 3 years ago)
- Last Synced: 2024-02-24T07:33:35.306Z (4 months ago)
- Language: Python
- Size: 6.33 MB
- Stars: 9
- Watchers: 9
- Forks: 4
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Lists
- awesome-self-driving-labs - Gemini - fidelity Bayesian optimization technique and is supported by Gryffin. (Software / Optimization)
README
# Gemini: Dynamic Bias Correction for Autonomous Experimentation and Molecular Simulation
[](https://travis-ci.com/rileyhickman/gemini)
[](https://codecov.io/gh/rileyhickman/gemini)
[](https://github.com/Naereen/StrapDown.js/blob/master/LICENSE)
Gemini is an open-source Python package which provides scalable multi-fidelity machine learning targeting
the design and discovery of functional molecules and advanced materials. (https://arxiv.org/abs/2103.03391v1)
## Installation
Install Gemini from source,
```bash
git clone https://github.com/rileyhickman/gemini.git
cd matter-gemini
pip install -e .
```
GPU use is optional. We recommend using the following
```bash
tensorflow-gpu 2.4.1
CUDA Version: 11.1
cuda-toolkit-11-1 11.1.1-1
Latest cuDNN
```
## Usage
### Supervised learning tasks with multi-fidelity data
Gemini can be easily trained given 2D (# samples, # dimensions) NumPy arrays containing
features (_x_) and targets (_y_) for _exp_ and _cheap_ datasets. Predictions using Gemini are
furnished with frequentist uncertainty estimates.
```python
from gemini import Gemini
gemini = Gemini()
gemini.train(x_exp, y_exp,
x_cheap, y_cheap)
pred_mu, pred_std = gemini.predict(x_exp_test)
```
### Scalable multi-fidelity Bayesian optimization
Gemini's predictions of expensive-to-evaluate objective functions can be used to
reduce the number of expensive black-box evaluations necessary to
achieve a desired target value.
The deep Bayesian optimizer Gryffin currently supports Gemini as a built-in predictive model.
After installing Gemini and Gryffin,
```python
from gryffin import Gryffin
# instantiate Gryffin
gryffin = Gryffin('config_file.json')
# optimization loop
while num_eval < budget:
samples = gryffin.recommend(observations,
proxy_observations)
```
The Gryffin config file must include a section specifying the predictive model, i.e.
```bash
...
"predictive_model": {
"model_kind": "gemini"
},
...
```
Alternatively, you can train Gemini in an external manner, this gives the user
greater flexibility in their expreiment. Gryffin allows for the optional passing of
a callable object to its `recommend` method.
```python
from gryffin import Gryffin
from gemini import GeminiOpt as Gemini
# instantiate Gryffin
gryffin = Gryffin('config_file.json')
# instantiate Gemini
gemini = Gemini()
# optimization loop
while num_eval < budget:
if len(observations) >= 2 and len(proxy_observations) >= 2:
# construct training set with current observations
training_set = gryffin.construct_training_set(observations, proxy_observations)
# train Gemini
gemini.train(training_set['train_features'], training_set['train_targets'],
training_set['proxy_train_features'], training_set['proxy_train_targets'],
num_folds=3)
# pass callable when asking Gryffin for new samples
samples = gryffin.recommend(observations,
predictive_model=gemini)
```
In this external trianing case, you need only provide a Gryffin config file (i.e. no predictive
model entries)
## Applications of Gemini (so far...)
* Inverse design of hybrid organic inorganic perovskites
* Inverse design of multi-component metal-oxide catalysts for the oxygen evolution reaction
* Inverse design of non-fullerene acceptor molecules for light harvesting applications
## Datasets
We provide methods for facile multi-fidelity data preprocessing/testing for 4 datasets reported in
the literature.
* `dataset_perovskites` (10.1038/sdata.2017.57)
* `dataset_freesolv` (10.1007/s10822-014-9747-x)
* `dataset_photobleaching` (10.1002/adma.201907801)
* `dataset_cat_oer_1_4` (10.1039/C9SC05999G)
## Contributing
Academic collaborations and extensions/improvements to the code are encouraged. Please reach out to Riley via email if you have questions/concerns.
## Developers
* Riley J. Hickman ([email protected])
* Florian Häse
* Matteo Aldeghi
## Citation
Gemini is an open-source research software. If you use Gemini in a scientific report, please cite the
following article
```
@misc{gemini,
title={Gemini: Dynamic Bias Correction for Autonomous Experimentation},
author={Riley J. Hickman and Florian Häse and Loïc M. Roch and Alán Aspuru-Guzik},
year={2021},
eprint={2103.03391},
archivePrefix={arXiv},
primaryClass={stat.ML}
}
```
## License
[MIT](https://choosealicense.com/licenses/mit/)