m53297601
/
easyAi
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

修改完成

Browse Source
pull/2/head
lidapeng 5 years ago
parent 5e0ed748d3
commit 3d2afcbb28
6 changed files with 52 additions and 21 deletions
Show Stats Download Patch File Download Diff File

32
src/main/java/org/wlld/imageRecognition/TempleConfig.java
Unescape View File

@ -26,9 +26,9 @@ import java.util.Map;
public class TempleConfig {
private NerveManager nerveManager;//神经网络管理器
private NerveManager convolutionNerveManager;//卷积神经网络管理器
private NerveManager convolutionNerveManagerR;//卷积神经网络管理器
private NerveManager convolutionNerveManagerG;//卷积神经网络管理器
private NerveManager convolutionNerveManagerB;//卷积神经网络管理器
private NerveManager convolutionNerveManagerR;//R卷积神经网络管理器
private NerveManager convolutionNerveManagerG;//G卷积神经网络管理器
private NerveManager convolutionNerveManagerB;//B卷积神经网络管理器
private boolean isAccurate = false;//是否保留精度
private int row = 5;//行的最小比例
private int column = 3;//列的最小比例
@ -48,6 +48,7 @@ public class TempleConfig {
private Normalization normalization = new Normalization();//统一归一化
private double avg = 0;//覆盖均值
private int sensoryNerveNub;//输入神经元个数
private boolean isShowLog = false;
public boolean isAccurate() {
return isAccurate;
@ -117,6 +118,27 @@ public class TempleConfig {
}
public void isShowLog(boolean isShowLog) {//是否打印学习数据
this.isShowLog = isShowLog;
}
public void startLvq() throws Exception {
switch (classifier) {
case Classifier.LVQ:
lvq.start();
break;
case Classifier.VAvg:
vectorK.study();
break;
}
if (isHavePosition) {
for (Map.Entry<Integer, KClustering> entry : kClusteringMap.entrySet()) {
entry.getValue().start();
}
boxReady = true;
}
}
private Map<Integer, KClustering> kClusteringMap = new HashMap<>();
public Map<Integer, KClustering> getKClusteringMap() {
@ -223,7 +245,7 @@ public class TempleConfig {
, int deep) throws Exception {
nerveManager = new NerveManager(sensoryNerveNub, 9,
classificationNub, deep, new Sigmod(), false, isAccurate);
nerveManager.init(initPower, false);
nerveManager.init(initPower, false, isShowLog);
}
private void initConvolutionVision(boolean initPower, int width, int height) throws Exception {//精准模式
@ -277,7 +299,7 @@ public class TempleConfig {
NerveManager convolutionNerveManager = new NerveManager(1, 1,
1, deep - 1, new ReLu(), true, isAccurate);
convolutionNerveManager.setMatrixMap(matrixMap);//给卷积网络管理器注入期望矩阵
convolutionNerveManager.init(initPower, true);
convolutionNerveManager.init(initPower, true, isShowLog);
return convolutionNerveManager;
}

7
src/main/java/org/wlld/nerveCenter/NerveManager.java
Unescape View File

@ -30,7 +30,7 @@ public class NerveManager {
private Map<Integer, Matrix> matrixMap = new HashMap<>();//主键与期望矩阵的映射
private boolean isDynamic;//是否是动态神经网络
private List<Double> studyList = new ArrayList<>();
private boolean isAccurate = false;//是否保留精度
private boolean isAccurate;//是否保留精度
public List<Double> getStudyList() {//查看每一次的学习率
return studyList;
@ -259,7 +259,7 @@ public class NerveManager {
* @param isMatrix 参数是否是一个矩阵
* @throws Exception
*/
public void init(boolean initPower, boolean isMatrix) throws Exception {//进行神经网络的初始化构建
public void init(boolean initPower, boolean isMatrix, boolean isShowLog) throws Exception {//进行神经网络的初始化构建
this.initPower = initPower;
initDepthNerve(isMatrix);//初始化深度隐层神经元
List<Nerve> nerveList = depthNerves.get(0);//第一层隐层神经元
@ -267,7 +267,8 @@ public class NerveManager {
List<Nerve> lastNeveList = depthNerves.get(depthNerves.size() - 1);
//初始化输出神经元
for (int i = 1; i < outNerveNub + 1; i++) {
OutNerve outNerve = new OutNerve(i, hiddenNerveNub, 0, studyPoint, initPower, activeFunction, isMatrix, isAccurate);
OutNerve outNerve = new OutNerve(i, hiddenNerveNub, 0, studyPoint, initPower,
activeFunction, isMatrix, isAccurate, isShowLog);
if (isMatrix) {//是卷积层神经网络
outNerve.setMatrixMap(matrixMap);
}

15
src/main/java/org/wlld/nerveEntity/OutNerve.java
Unescape View File

@ -19,10 +19,13 @@ import java.util.Map;
*/
public class OutNerve extends Nerve {
private Map<Integer, Matrix> matrixMapE;//主键与期望矩阵的映射
private boolean isShowLog;
public OutNerve(int id, int upNub, int downNub, double studyPoint, boolean init,
ActiveFunction activeFunction, boolean isDynamic, boolean isAccurate) throws Exception {
ActiveFunction activeFunction, boolean isDynamic, boolean isAccurate
, boolean isShowLog) throws Exception {
super(id, upNub, "OutNerve", downNub, studyPoint, init, activeFunction, isDynamic, isAccurate);
this.isShowLog = isShowLog;
}
@ -45,7 +48,9 @@ public class OutNerve extends Nerve {
} else {
this.E = 0;
}
System.out.println("E===" + this.E + ",out==" + out + ",nerveId==" + getId());
if (isShowLog) {
System.out.println("E==" + this.E + ",out==" + out + ",nerveId==" + getId());
}
gradient = outGradient();//当前梯度变化
//调整权重 修改阈值 并进行反向传播
updatePower(eventId);
@ -66,8 +71,10 @@ public class OutNerve extends Nerve {
Matrix myMatrix = dynamicNerve(matrix, eventId, isKernelStudy);
if (isKernelStudy) {//回传
Matrix matrix1 = matrixMapE.get(E);
System.out.println("E================" + E);
System.out.println(myMatrix.getString());
if (isShowLog) {
System.out.println("E================" + E);
System.out.println(myMatrix.getString());
}
if (matrix1.getX() <= myMatrix.getX() && matrix1.getY() <= myMatrix.getY()) {
double g = getGradient(myMatrix, matrix1);
backMatrix(g, eventId);

2
src/test/java/org/wlld/App.java
Unescape View File

@ -27,7 +27,7 @@ public class App {
public static void test3() throws Exception {
NerveManager nerveManager = new NerveManager(3, 6, 3
, 3, new Sigmod(), false, true);
nerveManager.init(true, false);//初始化
nerveManager.init(true, false, false);//初始化
List<Map<Integer, Double>> data = new ArrayList<>();//正样本
List<Map<Integer, Double>> dataB = new ArrayList<>();//负样本
List<Map<Integer, Double>> dataC = new ArrayList<>();//负样本

15
src/test/java/org/wlld/HelloWorld.java
Unescape View File

@ -125,8 +125,9 @@ public class HelloWorld {
// frame.setLengthWidth(640);
// templeConfig.setFrame(frame);
templeConfig.setClassifier(Classifier.DNN);
//templeConfig.isShowLog(true);
templeConfig.init(StudyPattern.Accuracy_Pattern, true, 640, 640, 2);
// ModelParameter modelParameter2 = JSON.parseObject(ModelData.DATA, ModelParameter.class);
// ModelParameter modelParameter2 = JSON.parseObject(ModelData.DATA2, ModelParameter.class);
// templeConfig.insertModel(modelParameter2);
Operation operation = new Operation(templeConfig);
//a b c d 物品 e是背景
@ -135,9 +136,9 @@ public class HelloWorld {
for (int i = 1; i < 1900; i++) {//一阶段
System.out.println("study1===================" + i);
//读取本地URL地址图片,并转化成矩阵
Matrix a = picture.getImageMatrixByLocal("D:\\share\\picture/a" + i + ".jpg");
Matrix a = picture.getImageMatrixByLocal("/Users/lidapeng/Desktop/myDocment/picture/a" + i + ".jpg");
//Matrix b = picture.getImageMatrixByLocal("/Users/lidapeng/Desktop/myDocment/picture/b" + i + ".jpg");
Matrix c = picture.getImageMatrixByLocal("D:\\share\\picture/c" + i + ".jpg");
Matrix c = picture.getImageMatrixByLocal("/Users/lidapeng/Desktop/myDocment/picture/c" + i + ".jpg");
//Matrix d = picture.getImageMatrixByLocal("/Users/lidapeng/Desktop/myDocment/picture/d" + i + ".jpg");
//Matrix f = picture.getImageMatrixByLocal("D:\\share\\picture/f" + i + ".png");
//将图像矩阵和标注加入进行学习，Accuracy_Pattern 模式 进行第二次学习
@ -149,8 +150,7 @@ public class HelloWorld {
//operation.learning(d, 4, false);
}
}
// ModelParameter modelParameter = JSON.parseObject(ModelData.DATA8, ModelParameter.class);
// templeConfig.insertModel(modelParameter);
//二阶段
for (int i = 1; i < 1900; i++) {
System.out.println("avg==" + i);
@ -194,8 +194,8 @@ public class HelloWorld {
// TempleConfig templeConfig2 = new TempleConfig(false);
// templeConfig2.init(StudyPattern.Accuracy_Pattern, true, 1000, 1000, 2);
// templeConfig2.insertModel(modelParameter2);
//
// Operation operation2 = new Operation(templeConfig2);
// Operation operation2 = new Operation(templeConfig2);
int wrong = 0;
int allNub = 0;
for (int i = 1900; i <= 2000; i++) {
@ -211,6 +211,7 @@ public class HelloWorld {
int cn = operation.toSee(c);
if (an != 1) {
wrong++;
} else {
}
if (cn != 2) {
wrong++;

2
src/test/java/org/wlld/NerveDemo1.java
Unescape View File

@ -33,7 +33,7 @@ public class NerveDemo1 {
* @param isDynamic 是否是动态神经元
*/
NerveManager nerveManager = new NerveManager(2, 6, 1, 4, new Sigmod(), false, true);
nerveManager.init(true, false);
nerveManager.init(true, false, false);
//创建训练

Write Preview
Loading…
Cancel
Save