# Install liblinear with Homebrew Library for large linear classification. Version 2.50 via Homebrew; verified from local package data. ## Install ```sh sudo av install brew:liblinear ``` Additional install commands: ### macOS - Homebrew (100%): ```sh brew install liblinear ``` Evidence: provider-native install command ## Package facts - **Package key:** brew:liblinear - **Package manager:** Homebrew - **Version:** 2.50 - **Source summary:** Library for large linear classification - **Homepage:** - **Repository:** - **Generated:** 2026-08-03T00:40:33+00:00 ## Executables - predict (alias) - train (alias) ## Install behavior - Bottle: not available ## Freshness - Page generated: 2026-08-03 - Package-manager version: 2.50 ## Project history and usage LIBLINEAR is a National Taiwan University machine-learning package for large-scale regularized linear classification, regression, and outlier detection. It pairs a C/C++ library with simple command-line tools such as train and predict, making linear SVM and logistic-regression models accessible from scripts and other software. ### Project history The LIBLINEAR paper was submitted to the Journal of Machine Learning Research in May 2008 and published in August 2008 by Rong-En Fan, Kai-Wei Chang, Cho-Jui Hsieh, Xiang-Rui Wang, and Chih-Jen Lin. The paper presents LIBLINEAR as an open source library for large-scale linear classification with easy command-line tools and library calls. LIBLINEAR was designed as a sibling to LIBSVM for cases where a linear model is appropriate and kernel methods are too expensive. The README explicitly advises beginners with small data sets to consider LIBSVM first, while pointing to LIBLINEAR for large data where nonlinear mappings do not materially improve performance. The public GitHub mirror was created on May 19, 2014, giving the project a familiar source-control home for packagers and downstream wrappers. ### Adoption history LIBLINEAR spread through both command-line use and embedding. Its stable file format, simple executables, MATLAB/OCTAVE interface, Python interface, and C API made it easy to package and wrap. One visible downstream adoption path is scikit-learn, whose LogisticRegression documentation exposes a 'liblinear' solver for L1 and L2 regularization. That made the project familiar even to Python users who never invoke the original train and predict binaries. ### How it is used The canonical workflow is to build the package with make, train a model from sparse feature data using train, and apply the model with predict. The README describes the LIBSVM-style data format, included heart_scale example, solver selection, cross validation, and parameter search. ### Why package nerds care LIBLINEAR is package-nerd catnip because it is a small, fast C/C++ implementation of a heavily cited ML method with stable Unix-style tools. It is the kind of dependency that appears behind higher-level language bindings while still remaining useful as a direct CLI package. ### Timeline - 2008: LIBLINEAR paper submitted in May and published in JMLR in August. - 2014: Public GitHub mirror created on May 19, 2014. - 2022: NTU-hosted LIBLINEAR paper PDF marked last modified on March 5, 2022. - 2025: GitHub repository metadata recorded push activity on December 29, 2025. ### Related projects - LIBSVM is the closest related project and the source of much of the package style, data-format familiarity, and user workflow. - scikit-learn exposes LIBLINEAR as a solver option in LogisticRegression, carrying the implementation into the Python scientific ecosystem. ### Sources - - - - - - ## Security Notes library-like package without higher-risk signals. - **Geiger risk:** green / low - library-like package without higher-risk signals ## Combined YAML source View the package source record on GitHub. [combined/liblinear.yml](https://github.com/automic-vault/db/blob/main/combined/liblinear.yml) ## Sources - Nucleus package database - Geiger risk classifier - curated package history - pkgdb category and tag curation