RFX-Fuse: Breiman and Cutler's Unified ML Engine + Native Explainable Similarity

RFX-Fuse: Breiman and Cutler's Random Forests as a Unified Machine Learning and Similarity Engine extended with Native Explainable Similarity. RFX-Fuse delivers Breiman and Cutler's complete vision for Random Forests as a unified machine learning and similarity engine with GPU/CPU acceleration. Breiman's original Random Forest paper (Breiman, 2001) and the Breiman-Cutler Fortran implementation included classification, regression, unsupervised learning, proximity-based similarity, outlier detection, and missing value imputation. Modern implementations like scikit-learn (2010) include only classification and regression. RFX-Fuse restores and extends the original Breiman and Cutler Random Forest capabilities. Novel Contribution: Native Explainable Similarity via Proximity Importance—answering "why are these samples similar?" directly from the model, without post-hoc computation. To our knowledge, this is the first similarity method with built-in explanations. Key Result: A single RFX-Fuse model provides a comparable alternative of up to 3-5 separate tools (XGBoost + SHAP + FAISS + Isolation Forest) across industry use cases: - Recommender Systems (1-2 models vs. 5 tools) - Finance Explainability (1 model vs. 3 tools) - Time Series Regression (1 model vs. 4 tools) - Imputation Quality Validation - Anomaly Detection