RFX-Fuse: A Unified Machine Learning Engine

RFX-Fuse (Random Forests X [X=compression] – Forest Unified Learning and Similarity Engine) aims to bring back Breiman and Cutler's original idea of Random Forests. The goal is to provide a complete machine learning engine that goes beyond simple ensemble prediction.

Features and Benefits

The library includes features such as classification, regression, unsupervised learning, proximity-based similarity, outlier detection, missing value imputation, and visualization. RFX-Fuse offers native support for GPUs and CPUs. It also introduces "Proximity Importance", a native explainable similarity that measures how similar samples are and explains why.

Simplifying Machine Learning Pipelines

Modern machine learning pipelines often require the use of several separate tools, such as XGBoost for prediction, FAISS for similarity, SHAP for explanations, and Isolation Forest for outlier detection. RFX-Fuse aims to consolidate these features into a single model, reducing the need for multiple tools and custom code.