Awesome Open Source
Awesome Open Source

Pyramid

A Java Machine Learning Library

Pyramid is a Java machine learning library which implements many state-of-the-art machine learning algorithms, including

At the moment, not all algorithms are released. We are actively working on tidying up the source files and adding documentations. We will release a few algorithms at a time when they are ready and hope to have all algorthms released soon!

Requirements

If you just want to use pyramid as a command line tool (which is very simple), all you need is Java 8.

If you are also a Java developer and wish to call Pyramid Java APIs, you will also need Maven.

Setup

Pyramid doesn't require any installation effort. All you need is downloading the latest pre-compiled package (with a name like pyramid-x.x.x.zip) and decompressing it. Now you can move into the created folder and type

./pyramid config/welcome.properties

You will see a welcome message and that means everything is working perfectly.

Windows users please see the notes.

Command Line Usage

All algorithms/functions implemented in Pyramid can be run though a simple command, with the following syntax:

./pyramid <properties_file>

Example:

./pyramid config/welcome.properties

or

./pyramid config/cbm.properties

pyramid is a launcher script and <properties_file> is a file specifying the name of the algorithm and all necessary parameters, such as the input data, output folder, and learning algorithm hyper parameters. The <properties_file> can be specified by either an absolute or a relative path.

To run different algorithms, you just need to invoke the program with different properties files. The list of available algorithms and their corresponding properties file templates can be found in the Wiki.

Building from Source

If you are a Java developer who prefer working with the source code or want to contribute to the Pyramid package:

Pyramid uses Maven for its build system.

To compile and package the project from the source code, simply run the mvn clean package -DskipTests command in the cloned directory. The compressed package will be created under the core/target/releases directory.

Feedback

We welcome your feedback on the package. To ask questions, request new features or report bugs, please contact Cheng Li via [email protected].

Answers to some commonly asked questions can be found in FAQ.


Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
Java (705,891
Machine Learning (30,060
Gradient Boosting (165
Multi Label Classification (129
Machine Learning Library (125
Related Projects