ShuJu

Framework for Advanced Statistics and Data Sciences

License	GNU GENERAL PUBLIC LICENSE, Version 3.0
GroupId	com.github.romualdrousseau
ArtifactId	shuju
Last Version	1.12
Release Date	Nov 3, 2020
Type	jar
Description	ShuJu Framework for Advanced Statistics and Data Sciences
Project URL	https://github.com/romualdrousseau/shuju
Source Code Management	https://github.com/romualdrousseau/shuju

Download shuju

Filename	Size
shuju-1.12.pom
shuju-1.12.jar	139 KB
shuju-1.12-sources.jar	69 KB
shuju-1.12-javadoc.jar	599 KB

How to add to project

Apache Maven

<!-- https://jarcasting.com/artifacts/com.github.romualdrousseau/shuju/ -->
<dependency>
    <groupId>com.github.romualdrousseau</groupId>
    <artifactId>shuju</artifactId>
    <version>1.12</version>
</dependency>

Gradle Groovy

// https://jarcasting.com/artifacts/com.github.romualdrousseau/shuju/
implementation 'com.github.romualdrousseau:shuju:1.12'

Gradle Kotlin

// https://jarcasting.com/artifacts/com.github.romualdrousseau/shuju/
implementation ("com.github.romualdrousseau:shuju:1.12")

Apache Buildr

'com.github.romualdrousseau:shuju:jar:1.12'

Apache Ivy

<dependency org="com.github.romualdrousseau" name="shuju" rev="1.12">
  <artifact name="shuju" type="jar" />
</dependency>

Groovy Grape

@Grapes(
@Grab(group='com.github.romualdrousseau', module='shuju', version='1.12')
)

Scala SBT

libraryDependencies += "com.github.romualdrousseau" % "shuju" % "1.12"

Leiningen

[com.github.romualdrousseau/shuju "1.12"]

Dependencies

compile (3)

Group / Artifact	Type	Version
org.processing : core	jar	3.3.7
com.fasterxml.jackson.core : jackson-databind	jar	2.10.0.pr1
org.xerial : sqlite-jdbc	jar	3.30.1

test (2)

Group / Artifact	Type	Version
junit : junit	jar	4.13.1
org.hamcrest : hamcrest-library	jar	1.3

Project Modules

There are no modules declared in this project.

ShuJu

It is an enterprise grade Java API for real linear algebra, statistic and machine learning algorithms.

Strongly influenced by Numpy and Tensorflow, ShuJu aims to provide a solid mathematical and scientific library for Java. Despite the success of python and various new languages, Java is still largely used in enterprise and mobile solutions because of its ubiquity and foundations in most of professional platforms.

Each implemented algorithm is not trivial to avoid for example calculation issues as rounding. Also a great care is given to optimization and tests.

The core of the linear algebra is (ala Numpy) an implementation of multi dimensional arrays with the notion of universal functions (lamda calculus). The implementation tries to avoid most of copies during tanspose, reshape and views and use packed java arrays to be friendly with CPU vectorization instructions or GPU. Fast computations is ensured by using BLAS subroutines (Experimental).

So far, the following algorithms have been implemented:

Math library

• Scalar
• Tensor (Including Vector, Matrix, ...)
• Linear Algebra algorithms
  • LU
  • QR
  • HouseGHolder
  • Hessenberg
  • Cholesky
  • Gaussian Elimination
  • Linear Solver
  • Eigens
  • PCA
  • SVD

Neural Network

• Full connected multiple layers (Dense)
• Convolution layers
• DropOut layers
• Batch Normlization Layers
• Flatten layers
• Pooling Layers
• Genetic Layers
• Gradient descent back propagation
• Activation functions (LeakyRelu, Linaer, Relu, Sigmoid, Softmax, Tanh)
• Loss function (Huber, Hinge, MeanSquare, SoftmaxCrossEntropy)
• Initializers (Glorot, He, Lecun)
• Regularizers (L1, L2)
• Optimizers (Adam, AdaDelta, RMSProp, Sgd)
• Schedulers (Exponential)

Machine Learnings

• Kmean
• K-Nearest Neighbor
• Naive Bayes
• Simple Linear Regression

NLP

• Tokenizers (ngram, shingle)
• Word 2 vect
• Regular Expression
• Stop words

CV

• Image convolution filter
• Shape extractors (rectangle by hough transformation)

Genetic

• Pool with selection
• Mutation interface

DataSet management

With transformation framework such as normalization, standardization ...

Project Documentation

https://romualdrousseau.github.io/ShuJu/

User Guide

https://github.com/RomualdRousseau/ShuJu/wiki

Install Processing examples

• Set environement variable PROCESSING_USER_LIBRARIES to the Processing User libraries
• mvn clean install

Manual Deployment

Manually Deploy snapshot

mvn clean deploy

Manually Deploy release

mvn clean site -Dmessage="<commit_log_here>" scm:checkin -P release
mvn scm:update

Automatic Deployment

Deploy snapshot

Every time a change on master is made (push, merge, etc) a deployment on the snapshot repository is made.

Deploy release

• Edit the pom.xml to set the release version and push
• Create a release on github will trigger a deployment on the official repo as well as generate the associated javadoc.
• Edit the pom.xml to set the snapshot version and push

Deploy documentation

mvn clean site deploy -P documentation

Versions

Version
1.12 Nov 3, 2020
1.11 Oct 30, 2020
1.10 Oct 7, 2020
1.9 Sep 17, 2020
1.8 Jul 14, 2020
1.7 Jul 14, 2020
1.6 Feb 12, 2020
1.5 Nov 11, 2018
1.4 Sep 21, 2018
1.3 Jul 6, 2018
1.2 Jun 15, 2018
1.1 Jun 15, 2018
1.0 Jun 15, 2018