- 切换到 中文文档
- knowledge base
The AS machine learning library provides a Java API for various machine learning algorithms. It has the encapsulation of complex computing logic such as machine vision and natural language processing. Through the library, various algorithms can be quickly used to achieve various effects. Many calculation operations in the AS library are implemented natively, enabling the operation of library functions without unexpected dependencies from the Java standard library.
There are multiple versions of the AS library directory. If you want to query the update logs and differences of different versions, you can view here.
A test dataset is provided in this warehouse, which contains various data files such as images. You can pull these test data into a local program through a URL for calculation.
⚠️ 【 Important 】 The content of versions 1.32 and 1.40 is almost identical, with the main difference being the changes to the package module. Please note that we will refactor the package name toio. github. beardedManZhao. algorithmStarin versions 1.40 and all subsequent versions to avoid conflicting package names in many Java dependencies~To avoid any compatibility issues caused by package updates, we have provided version 1.32. You can continue to use the old package name, but we strongly recommend using the new version, as the new package name has been updated to 'io. github. beardedManZhao. algorithmStar'. If you have any questions or suggestions about changing the package name or updating, please contact us in a timely manner!!
You can integrate Arithmetic Star into your project through maven, and the configuration of Maven is shown below. You can add it to your maven project, or you can download it from Releases and manually integrate it into your project.
<dependencies>
<dependency>
<groupId>io.github.BeardedManZhao</groupId>
<artifactId>algorithmStar</artifactId>
<version>1.40</version>
</dependency>
</dependencies>After version 1.17, all dependencies of the AS library have been stripped to better avoid binding dependencies and reduce the possibility of project conflicts. At the same time, more suitable dependency configuration items can be used according to the needs of developers. You can view third-party library dependencies on which the AS library depends here.
The AS library generates some log data when performing many calculation functions. Therefore, the use of the AS library requires importing log dependencies, which are essential. Please import the dependencies as follows.
<dependencies>
<!-- Binding using the adapter of log4j2 -->
<dependency>
<groupId>org.apache.logging.log4j</groupId>
<artifactId>log4j-slf4j-impl</artifactId>
<version>2.20.0</version>
<!--<scope>provided</scope>-->
</dependency>
<!-- Log4j2 log facade -->
<dependency>
<groupId>org.apache.logging.log4j</groupId>
<artifactId>log4j-api</artifactId>
<version>2.20.0</version>
<!--<scope>provided</scope>-->
</dependency>
<!-- Log4j2 log real surface -->
<dependency>
<groupId>org.apache.logging.log4j</groupId>
<artifactId>log4j-core</artifactId>
<version>2.20.0</version>
<!--<scope>provided</scope>-->
</dependency>
</dependencies>When interfacing with various platforms such as databases and Sparks, the AS library needs to use third-party dependency packages, which are optional. If you do not need to use these functions, you may not need to import dependencies. If you need to, you can refer to the following configuration.
<dependencies>
<!-- MySQL database connection driver If the relational database you want to connect to is of another type, you can also modify it here -->
<dependency>
<groupId>mysql</groupId>
<artifactId>mysql-connector-java</artifactId>
<version>8.0.30</version>
</dependency>
<!-- The dependency development package for the three major Spark modules can also be imported if you need to use it here, or not if you don't need it -->
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-core_2.12</artifactId>
<version>3.4.0</version>
</dependency>
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-sql_2.12</artifactId>
<version>3.4.0</version>
</dependency>
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-mllib_2.12</artifactId>
<version>3.4.0</version>
</dependency>
<!-- Camera device dependency library. If you have a need to obtain data objects through the camera, you can import this dependency. -->
<dependency>
<groupId>com.github.sarxos</groupId>
<artifactId>webcam-capture</artifactId>
<version>0.3.12</version>
</dependency>
<!-- HDFS input/output device dependency library. If you have a need for data reading and writing through the HDFS distributed storage platform, you can introduce this library. -->
<dependency>
<groupId>org.apache.hadoop</groupId>
<artifactId>hadoop-client</artifactId>
<version>3.3.1</version>
</dependency>
</dependencies>After loading the AS library, you can use the following code to download all the help documents to your local computer. There are some sample codes to guide you in using them, and you can make modifications according to your own needs.
package io.github.beardedManZhao.algorithmStar;
import io.github.beardedManZhao.algorithmStar.algorithm.distanceAlgorithm.EuclideanMetric;
import io.github.beardedManZhao.algorithmStar.algorithm.distanceAlgorithm.ManhattanDistance;
import io.github.beardedManZhao.algorithmStar.algorithm.featureExtraction.WordFrequency;
import io.github.beardedManZhao.algorithmStar.core.AlgorithmStar;
import io.github.beardedManZhao.algorithmStar.operands.matrix.ColumnIntegerMatrix;
public final class MAIN1 {
public static void main(String[] args) {
// 准备两个文本组成的数组，其中是两个需要被处理成为特征向量的字符串语句。
String[] data = {
"Good evening, dear, don't forget the agreement between us. It's 9:00 tomorrow morning.",
"Good morning, dear, don't forget the agreement between us, at 9:00 in the morning."
};
AlgorithmStar<Object, ColumnIntegerMatrix> algorithmStar = AlgorithmStar.getInstance();
// 开始进行特征提取，生成词频向量 提取成功之后会返回一个矩阵
ColumnIntegerMatrix word = algorithmStar.extract(WordFrequency.getInstance("word"), data);
// 打印生成的向量矩阵
System.out.println(word);
// 开始将矩阵中的两句话对应的向量获取到
int[] arrayByColIndex1 = word.getArrayByColIndex(0);
int[] arrayByRowIndex2 = word.getArrayByColIndex(1);
// 开始计算两个向量之间的德式距离
double res = algorithmStar.getTrueDistance(EuclideanMetric.getInstance("e"), arrayByColIndex1, arrayByRowIndex2);
System.out.println(res);
// 开始计算两个向量之间的曼哈顿距离
double res1 = algorithmStar.getTrueDistance(ManhattanDistance.getInstance("man"), arrayByColIndex1, arrayByRowIndex2);
System.out.println(res1);
AlgorithmStar.clear();
}
}Feature computing components are often used in feature engineering, and are the main force of computing in machine learning tasks. Measurement algorithms, classification algorithms, etc. can use this component computing method. The feature engineering calculation in Arithmetic Star is component-based calculation. You can directly obtain component objects and call component functions, or use Arithmetic Star's portal class for feature engineering.
Next is a simple example of feature engineering calculation using algorithmic star portal class
import io.github.beardedManZhao.algorithmStar.integrator.Route2DDrawingIntegrator;
import io.github.beardedManZhao.algorithmStar.algorithm.generatingAlgorithm.ZhaoCoordinateNet2D;
import io.github.beardedManZhao.algorithmStar.operands.coordinate.DoubleCoordinateTwo;
import io.github.beardedManZhao.algorithmStar.operands.route.DoubleConsanguinityRoute2D;
/**
* 示例代码文件
*/
public class MAIN1 {
public static void main(String[] args) {
// 构建人员坐标(二维)
DoubleCoordinateTwo A = new DoubleCoordinateTwo(10, 10);
DoubleCoordinateTwo B = new DoubleCoordinateTwo(-10, 4);
DoubleCoordinateTwo C = new DoubleCoordinateTwo(1, 0);
DoubleCoordinateTwo E = new DoubleCoordinateTwo(6, 1);
DoubleCoordinateTwo Z = new DoubleCoordinateTwo(1, 21);
// 获取关系网络,该算法是我实现出来,用于推断人员关系网的,这里的名称您可以自定义,需要注意的是下面集成器的实例化需要您将该名称传进去
ZhaoCoordinateNet2D zhaoCoordinateNet = ZhaoCoordinateNet2D.getInstance("Z");
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("A -> B", A, B)); // Representing A takes the initiative to know B
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("A -> C", A, C));
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("E -> Z", E, Z));
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("A -> Z", A, Z));
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("B -> Z", B, Z));
// 使用2维路线绘制集成器,输出上面所有人员之间的关系网络图片
Route2DDrawingIntegrator a = new Route2DDrawingIntegrator("A", "Z");
// 设置图片输出路径
a.setImageOutPath("D:\\out\\image.jpg")
// 设置图片宽度
.setImageWidth(1000)
// 设置图片高度
.setImageHeight(1000)
// 设置离散阈值,用来放大微小的变化
.setDiscreteThreshold(4)
// 运行集成器!
.run();
// 清理关系网络中的数据
zhaoCoordinateNet.clear();
}
}- run results
[INFO][OperationAlgorithmManager][23-01-18:01]] : register OperationAlgorithm:word
------------IntegerMatrixStart-----------
Good evening, dear, don't forget the agreement between us. It's 9:00 tomorrow morning. Good morning, dear, don't forget the agreement between us, at 9:00 in the morning. rowColName
[1, 1] 00
[1, 1] agreement
[1, 1] don't
[0, 1] in
[1, 0] tomorrow
[1, 2] morning
[1, 2] the
[1, 1] forget
[0, 1] at
[1, 0] It's
[1, 1] 9
[1, 1] Good
[1, 0] evening
[1, 1] dear
[1, 1] between
[1, 1] us
------------IntegerMatrixEnd------------
[INFO][OperationAlgorithmManager][23-01-18:01]] : register OperationAlgorithm:e
2.6457513110645907
[INFO][OperationAlgorithmManager][23-01-18:01]] : register OperationAlgorithm:man
7.0
进程已结束,退出代码0
Here is an example of the use of "ZhaoCoordinateNet2D", which shows the analysis and prediction of the relationship between people. In the "ZhaoCoordinateNet" algorithm, the relationship network of people is analyzed, and at the same time, the network analysis and mapping can be performed.
package zhao.algorithmMagic;
import zhao.algorithmMagic.lntegrator.Route2DDrawingIntegrator;
import zhao.algorithmMagic.algorithm.generatingAlgorithm.ZhaoCoordinateNet2D;
import zhao.algorithmMagic.operands.coordinate.DoubleCoordinateTwo;
import zhao.algorithmMagic.operands.route.DoubleConsanguinityRoute2D;
/**
* 示例代码文件
*/
public class MAIN1 {
public static void main(String[] args) {
// Builder coordinates (2D)
DoubleCoordinateTwo A = new DoubleCoordinateTwo(10, 10);
DoubleCoordinateTwo B = new DoubleCoordinateTwo(-10, 4);
DoubleCoordinateTwo C = new DoubleCoordinateTwo(1, 0);
DoubleCoordinateTwo E = new DoubleCoordinateTwo(6, 1);
DoubleCoordinateTwo Z = new DoubleCoordinateTwo(1, 21);
/*
Get the relationship network. This algorithm is implemented by me to infer the relationship network of people.
You can customize the name here. It should be noted that the instantiation of the integrator below requires you to pass the name in.
*/
ZhaoCoordinateNet2D zhaoCoordinateNet = ZhaoCoordinateNet2D.getInstance("Z");
/*
Add the relationship of people to the relationship network, please note that the relationship network of the algorithm already contains your data,
so you must pass the same name in the integration below, so that the integrator can get the temporary network in your algorithm grid data
*/
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("A -> B", A, B)); // Representing A takes the initiative to know B
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("A -> C", A, C));
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("E -> Z", E, Z));
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("A -> Z", A, Z));
zhaoCoordinateNet.addRoute(DoubleConsanguinityRoute2D.parse("B -> Z", B, Z));
// Use a 2-dimensional route drawing integrator to output a picture of the relationship network between all the people above
Route2DDrawingIntegrator a = new Route2DDrawingIntegrator("A", "Z");
// Set image output path
a.setImageOutPath("D:\\out\\image.jpg")
// set image width
.setImageWidth(1000)
// set image height
.setImageHeight(1000)
// Set discrete thresholds to amplify small changes
.setDiscreteThreshold(4)
// Run the integrator!
.run();
// Clean data in relational network
zhaoCoordinateNet.clear();
}
}-
The relationship network picture generated after running
The directory contains the DLL dynamic library file and the source code corresponding to the library file that the Algorithm Star machine learning library depends on when it runs. In order to achieve better compatibility in the machine learning library, the C compiled DLL is introduced. Users can load the DLL into the machine learning library before the AS library runs.
The knowledge base file archive is used for the storage task of AS-MB series knowledge documents. You can access it directly through the documents on the home page. There is no need to enter here directly. It is messy internally.
The source code storage directory of AS-MB, where you can view the relevant source code of AS-MB. Of course, here is the latest source code, which you can use to compile, so that you can obtain the latest version.
Note: The latest version is often unstable, so we recommend you to use the version that has been released for a long time!
The Chinese version of the AS-MB homepage document. You can switch languages on the homepage default page to access this file.
The default version of the AS-MB homepage document. You can directly access this file on the homepage!
- 切换到 中文文档