ralfbiedert/ffsvm — 77 Stars

Topics

#libsvm #machine-learning #rust #svm-library

Language breakdown

No language data available.

▸ Files click to expand

README

[![crates.io-badge]][crates.io-url] [![docs.rs-badge]][docs.rs-url] ![license-badge] [![rust-version-badge]][rust-version-url] [![rust-build-badge]][rust-build-url]

In One Sentence

You trained an SVM using libSVM, now you want the highest possible performance during (real-time) classification, like games or VR.

Highlights

loads almost all libSVM types (C-SVC, ν-SVC, ε-SVR, ν-SVR) and kernels (linear, poly, RBF and sigmoid)
produces practically same classification results as libSVM
optimized for SIMD and can be mixed seamlessly with Rayon
written in 100% safe Rust
allocation-free during classification for dense SVMs
2.5x - 14x faster than libSVM for dense SVMs
extremely low classification times for small models (e.g., 128 SV, 16 dense attributes, linear ~ 500ns)
successfully used in Unity and VR projects (Windows & Android)

Usage

Train with libSVM (e.g., using the tool svm-train), then classify with ffsvm-rust.

From Rust:

// Replace SAMPLE_MODEL with a &amp;str to your model.
let svm = DenseSVM::tryfrom(SAMPLEMODEL)?;
let mut fv = FeatureVector::from(&svm); let features = fv.features();
features[0] = 0.55838; features[1] = -0.157895; features[2] = 0.581292; features[3] = -0.221184;
svm.predict_value(&mut fv)?;
assert_eq!(fv.label(), Label::Class(42));

Status

December 14, 2024: After 7+ years, finally ported to stable.^🎉^🎉^🎉
March 10, 2023: Reactivated for latest Rust nightly.
June 7, 2019: Gave up on 'no unsafe', but gained runtime SIMD selection.
March 10, 2019: As soon as we can move away from nightly we'll go beta.
Aug 5, 2018: Still in alpha, but finally on crates.io.
May 27, 2018: We're in alpha. Successfully used internally on Windows, Mac, Android and Linux

on various machines and devices. Once SIMD stabilizes and we can cross-compile to WASM we'll move to beta.

December 16, 2017: We're in pre-alpha. It will probably not even work on your machine.

Performance

performance

All performance numbers reported for the DenseSVM. We also have support for SparseSVMs, which are slower for "mostly dense" models, and faster for "mostly sparse" models (and generally on the performance level of libSVM).

See here for details.

Tips

Compile your project with target-cpu=native for a massive speed boost (e.g., check our .cargo/config.toml how

you can easily do that for your project). Note, due to how Rust works, this is only used for application (or dynamic FFI libraries), not library crates wrapping us.

For an x-fold performance increase, create a number of Problem structures, and process them with Rayon's pariter.

FAQ

See here for details.

[crates.io-badge]: https://img.shields.io/crates/v/ffsvm.svg [crates.io-url]: https://crates.io/crates/ffsvm [license-badge]: https://img.shields.io/badge/license-MIT-blue.svg [docs.rs-badge]: https://docs.rs/ffsvm/badge.svg [docs.rs-url]: https://docs.rs/ffsvm/ [rust-version-badge]: https://img.shields.io/badge/rust-1.83%2B-blue.svg?maxAge=3600 [rust-version-url]: https://github.com/ralfbiedert/ffsvm [rust-build-badge]: https://github.com/ralfbiedert/ffsvm/actions/workflows/rust.yml/badge.svg [rust-build-url]: https://github.com/ralfbiedert/ffsvm/actions/workflows/rust.yml

🔗 More in this category