Learned-Index-Benefits (LIB) is the implementation of the paper -- Learned Index Benefits: Machine Learning Based Index Performance Estimation. It is an end-to-end machine learning based index benefit estimator. The objective of this model is to facilitate index selection process by accurately and efficiently quantifying the benefits of index configuration on a query.
- Python 3.8
- Numpy
- Torch
- Sklearn
The Test dataset is placed in /data/ directory. The data is generated according to the techniques discussed in the paper and stored as a tuple <vector representation, cost reduction ratio>.
The Pytorch implementation of LIB is shown in the notebook above.