Panern/SimBoOpt

A simple Bayesian Optimization framework for ML and DL.

SimBoOpt

A simple Bayesian Optimization framework for ML and DL, which can be used for fine-tuning parameters of Models efficently. Generally, your can obtain better performance of your model.

How we use it?

Needed:

Define your model

do this in Model.py.

Setting paras of experiments

do this in template.yaml.

Make sure BoOp.py is modified depended on the former changes.

If Needed:

Define your data function

do this in DataProcess.py

Define your evaluation function

do this TrainAndEvaluate

Examples

We have two examples of SimBoOpt:

1. DL: CNN for classifaction of Minist
2. ML: MAGC-based(paper: Multi-view Attributed Graph Clustering for attributed graph clustering on ACM datasets

Dependencies

1. Numpy
2. Scipy
3. Scikit-learn
4. Ax
5. Scanpy
6. Torch
7. Traceback
8. Bayes_opt

More details can be seen in code comment.

Valuable contribution for this repository will be accepted.

Cite this repository if U follow our works.