/linfa

A Rust machine learning framework.

Primary LanguageRustApache License 2.0Apache-2.0

Linfa mascot icon

Linfa

crates.io Documentation DocumentationLatest Codequality Run Tests

linfa (Italian) / sap (English):

The vital circulating fluid of a plant.

linfa aims to provide a comprehensive toolkit to build Machine Learning applications with Rust.

Kin in spirit to Python's scikit-learn, it focuses on common preprocessing tasks and classical ML algorithms for your everyday ML tasks.

Website | Community chat

Current state

Where does linfa stand right now? Are we learning yet?

linfa currently provides sub-packages with the following algorithms:

Name Purpose Status Category Notes
clustering Data clustering Tested / Benchmarked Unsupervised learning Clustering of unlabeled data; contains K-Means, Gaussian-Mixture-Model, DBSCAN and OPTICS
kernel Kernel methods for data transformation Tested Pre-processing Maps feature vector into higher-dimensional space
linear Linear regression Tested Partial fit Contains Ordinary Least Squares (OLS), Generalized Linear Models (GLM)
elasticnet Elastic Net Tested Supervised learning Linear regression with elastic net constraints
logistic Logistic regression Tested Partial fit Builds two-class logistic regression models
reduction Dimensionality reduction Tested Pre-processing Diffusion mapping and Principal Component Analysis (PCA)
trees Decision trees Tested / Benchmarked Supervised learning Linear decision trees
svm Support Vector Machines Tested Supervised learning Classification or regression analysis of labeled datasets
hierarchical Agglomerative hierarchical clustering Tested Unsupervised learning Cluster and build hierarchy of clusters
bayes Naive Bayes Tested Supervised learning Contains Gaussian Naive Bayes
ica Independent component analysis Tested Unsupervised learning Contains FastICA implementation
pls Partial Least Squares Tested Supervised learning Contains PLS estimators for dimensionality reduction and regression
tsne Dimensionality reduction Tested Unsupervised learning Contains exact solution and Barnes-Hut approximation t-SNE
preprocessing Normalization & Vectorization Tested / Benchmarked Pre-processing Contains data normalization/whitening and count vectorization/tf-idf
nn Nearest Neighbours & Distances Tested / Benchmarked Pre-processing Spatial index structures and distance functions
ftrl Follow The Reguralized Leader - proximal Tested / Benchmarked Partial fit Contains L1 and L2 regularization. Possible incremental update

We believe that only a significant community effort can nurture, build, and sustain a machine learning ecosystem in Rust - there is no other way forward.

If this strikes a chord with you, please take a look at the roadmap and get involved!

BLAS/Lapack backend

Some algorithm crates need to use an external library for linear algebra routines. By default, we use a pure-Rust implementation. However, you can also choose an external BLAS/LAPACK backend library instead, by enabling the blas feature and a feature corresponding to your BLAS backend. Currently you can choose between the following BLAS/LAPACK backends: openblas, netblas or intel-mkl.

Backend Linux Windows macOS
OpenBLAS ✔️ - -
Netlib ✔️ - -
Intel MKL ✔️ ✔️ ✔️

Each BLAS backend has two features available. The feature allows you to choose between linking the BLAS library in your system or statically building the library. For example, the features for the intel-mkl backend are intel-mkl-static and intel-mkl-system.

An example set of Cargo flags for enabling the Intel MKL backend on an algorithm crate is --features blas,linfa/intel-mkl-system. Note that the BLAS backend features are defined on the linfa crate, and should only be specified for the final executable.

License

Dual-licensed to be compatible with the Rust project.

Licensed under the Apache License, Version 2.0 http://www.apache.org/licenses/LICENSE-2.0 or the MIT license http://opensource.org/licenses/MIT, at your option. This file may not be copied, modified, or distributed except according to those terms.