KMeans (LIRE API)

跳过导航链接

所有类

概要:
嵌套 |
字段 |
构造器 |
方法

详细资料:
字段 |
构造器 |
方法

java.lang.Object
- net.semanticmetadata.lire.classifiers.KMeans

直接已知子类:

ParallelKMeans
```
public class KMeans
extends java.lang.Object
```
Created by Mathias on 23/09/2008.

作者:

Mathias Lux, mathias@juggle.at, Nektarios Anagnostopoulos, nek.anag@gmail.com, Lazaros Tsochatzidis, ltsochat@ee.duth.gr

字段概要

字段
限定符和类型	字段和说明
`protected Cluster[]`	`clusters`
`protected int`	`countAllFeatures`
`protected java.util.ArrayList<double[]>`	`features`
`protected int`	`length`
`protected int`	`numClusters`

构造器概要

构造器
构造器和说明

KMeans(int numClusters)

方法概要

所有方法实例方法具体方法
限定符和类型	方法和说明
`void`	`addFeature(double[] feature)`
`double`	`clusteringStep()` Do one step and return the overall stress (squared error).
`Cluster[]`	`getClusters()`
`int`	`getFeatureCount()`
`int`	`getNumClusters()` Get the number of desired clusters.
`protected boolean`	`hasNaNs(double[] histogram)`
`void`	`init()`
`protected double`	`overallStress()` Squared error in classification.
`protected void`	`recomputeMeans()` Computes the mean per cluster (averaged vector)
`protected void`	`reOrganizeFeatures()` Re-shuffle all features.
`protected java.util.Set<java.lang.Integer>`	`selectInitialMedians(int numClusters)`

从类继承的方法 java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

字段详细资料

countAllFeatures
```
protected int countAllFeatures
```

numClusters
```
protected int numClusters
```

length
```
protected int length
```

features

protected java.util.ArrayList<double[]> features

clusters
```
protected Cluster[] clusters
```

构造器详细资料
- KMeans
```
public KMeans(int numClusters)
```

方法详细资料
- addFeature
```
public void addFeature(double[] feature)
```
- init
```
public void init()
```
- selectInitialMedians
```
protected java.util.Set<java.lang.Integer> selectInitialMedians(int numClusters)
```
- clusteringStep
```
public double clusteringStep()
```
  Do one step and return the overall stress (squared error). You should do this until the error is below a threshold or doesn't change a lot in between two subsequent steps.
  
  返回:
- hasNaNs
```
protected boolean hasNaNs(double[] histogram)
```
- reOrganizeFeatures
```
protected void reOrganizeFeatures()
```
  Re-shuffle all features.
- recomputeMeans
```
protected void recomputeMeans()
```
  Computes the mean per cluster (averaged vector)
- overallStress
```
protected double overallStress()
```
  Squared error in classification.
  
  返回:
- getNumClusters
```
public int getNumClusters()
```
  Get the number of desired clusters.
  
  返回:
- getFeatureCount
```
public int getFeatureCount()
```
- getClusters
```
public Cluster[] getClusters()
```

跳过导航链接

所有类

概要:
嵌套 |
字段 |
构造器 |
方法

详细资料:
字段 |
构造器 |
方法