This is really a cool work out of Microsoft research called hummingbird. You can convert traditional machine learning models to tensor computations to take advantage of hardware acceleration like GPUs and TPUs. It allows users to seamlessly leverage neural network frameworks (such as PyTorch) to accelerate traditional ML models.
It has many features and benefits as follows:
- User can benefit from current and future optimizations implemented in neural network frameworks;
- User can benefit from native hardware acceleration;
- User can benefit from having a unique platform to support both traditional and neural network models;
- User does not have to re-engineer their models.
Hummingbird is compatible with a number of tree-based classifiers and regressors. These models include scikit-learn Decision Trees and Random Forest.
Here they convert a random forest model to PyTorch (https://github.com/microsoft/hummingbird)