Crate rustkernel_ml 

RustKernel Statistical ML

GPU-accelerated machine learning kernels for clustering, anomaly detection, and regression.

Kernels

Clustering (3 kernels)

• KMeans - Lloyd’s algorithm with K-Means++ initialization
• DBSCAN - Density-based clustering with GPU union-find
• HierarchicalClustering - Agglomerative clustering

Anomaly Detection (2 kernels)

• IsolationForest - Ensemble of isolation trees
• LocalOutlierFactor - k-NN density estimation

Streaming Anomaly Detection (2 kernels)

• StreamingIsolationForest - Online anomaly detection with sliding window
• AdaptiveThreshold - Self-adjusting thresholds with drift detection

Ensemble (1 kernel)

• EnsembleVoting - Weighted majority voting

Regression (2 kernels)

• LinearRegression - OLS via normal equations
• RidgeRegression - L2 regularization

Explainability (2 kernels)

• SHAPValues - Kernel SHAP for feature explanations
• FeatureImportance - Permutation-based feature importance

NLP / LLM Integration (2 kernels)

• EmbeddingGeneration - Text embedding via hash-based features
• SemanticSimilarity - Multi-metric semantic similarity

Federated Learning (1 kernel)

• SecureAggregation - Privacy-preserving model aggregation

Healthcare Analytics (2 kernels)

• DrugInteractionPrediction - Multi-drug interaction prediction
• ClinicalPathwayConformance - Treatment guideline checking

Modules

anomaly

Anomaly detection kernels.

clustering

Clustering kernels.

ensemble

Ensemble method kernels.

explainability

Explainability kernels for model interpretation.

federated

Federated Learning kernels.

healthcare

Healthcare analytics kernels.

messages

Ring message types for ML kernels.

nlp

Natural Language Processing and LLM integration kernels.

prelude

Prelude for convenient imports.

regression

Regression kernels.

ring_messages

Ring message types for Statistical ML kernels.

streaming

Streaming anomaly detection kernels.

types

Common ML types and data structures.

Functions

register_all

Register all ML kernels with a registry.