From 7f7841c5c4db35faa75a1ab7f755e7ae76fa2662 Mon Sep 17 00:00:00 2001 From: lidapeng Date: Sun, 12 Jan 2020 21:42:55 +0800 Subject: [PATCH] =?UTF-8?q?=E6=8F=90=E9=80=9F=E9=98=89=E5=89=B2=E7=89=88?= =?UTF-8?q?=E5=8D=B7=E7=A7=AF?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- src/main/java/org/wlld/App.java | 15 +-- .../wlld/imageRecognition/Convolution.java | 2 +- .../org/wlld/imageRecognition/Operation.java | 17 +-- .../wlld/imageRecognition/TempleConfig.java | 24 ---- .../org/wlld/nerveCenter/NerveManager.java | 5 +- .../org/wlld/nerveEntity/HiddenNerve.java | 18 +-- src/main/java/org/wlld/nerveEntity/Nerve.java | 114 +++--------------- .../java/org/wlld/nerveEntity/OutNerve.java | 22 +--- .../org/wlld/nerveEntity/SensoryNerve.java | 6 +- src/main/java/org/wlld/test/Test.java | 7 +- 10 files changed, 34 insertions(+), 196 deletions(-) diff --git a/src/main/java/org/wlld/App.java b/src/main/java/org/wlld/App.java index 1391f17..67906ae 100644 --- a/src/main/java/org/wlld/App.java +++ b/src/main/java/org/wlld/App.java @@ -18,8 +18,8 @@ import java.util.Map; public class App { public static void main(String[] args) throws Exception { //createNerveTest(); - //testPic(); - test(); + testPic(); + //test(); } public static void testPic() throws Exception { @@ -34,7 +34,7 @@ public class App { Map wrongTagging = new HashMap<>();//分类标注 rightTagging.put(1, 1.0); wrongTagging.put(1, 0.0); - for (int i = 1; i < 1000; i++) { + for (int i = 1; i < 999; i++) { System.out.println("开始学习1==" + i); //读取本地URL地址图片,并转化成矩阵 Matrix right = picture.getImageMatrixByLocal("/Users/lidapeng/Desktop/myDocment/c/c" + i + ".png"); @@ -51,9 +51,9 @@ public class App { ModelParameter modelParameter = templeConfig.getModel(); String model = JSON.toJSONString(modelParameter); System.out.println("一次提取：" + model); + ModelParameter modelParameter1 = JSONObject.parseObject(model, ModelParameter.class); TempleConfig templeConfig1 = getTemple(false); - ModelParameter modelParameter1 = JSONObject.parseObject(model, ModelParameter.class); templeConfig1.insertModel(modelParameter1); ModelParameter modelParameter2 = templeConfig1.getModel(); String model2 = JSON.toJSONString(modelParameter2); @@ -63,13 +63,8 @@ public class App { Matrix right = picture.getImageMatrixByLocal("/Users/lidapeng/Desktop/myDocment/test/a101.png"); Matrix wrong = picture.getImageMatrixByLocal("/Users/lidapeng/Desktop/myDocment/b/b1000.png"); //进行图像检测 - operation1.look(right, 2); operation1.look(wrong, 3); - } - - public static void test() throws Exception { - TempleConfig templeConfig = new TempleConfig(); - templeConfig.initConvolutionalVision(true, 3024, 4032); + operation1.look(right, 2); } public static TempleConfig getTemple(boolean isFirst) throws Exception { diff --git a/src/main/java/org/wlld/imageRecognition/Convolution.java b/src/main/java/org/wlld/imageRecognition/Convolution.java index 094c133..642a953 100644 --- a/src/main/java/org/wlld/imageRecognition/Convolution.java +++ b/src/main/java/org/wlld/imageRecognition/Convolution.java @@ -3,6 +3,7 @@ package org.wlld.imageRecognition; import org.wlld.MatrixTools.Matrix; import org.wlld.MatrixTools.MatrixOperation; import org.wlld.config.Kernel; +import org.wlld.tools.ArithUtil; /** * @author lidapeng @@ -33,7 +34,6 @@ public class Convolution { } return late(myMatrix); } - public Matrix late(Matrix matrix) throws Exception {//迟化处理 int xn = matrix.getX(); int yn = matrix.getY(); diff --git a/src/main/java/org/wlld/imageRecognition/Operation.java b/src/main/java/org/wlld/imageRecognition/Operation.java index 0e7e258..bb2a2bb 100644 --- a/src/main/java/org/wlld/imageRecognition/Operation.java +++ b/src/main/java/org/wlld/imageRecognition/Operation.java @@ -2,6 +2,7 @@ package org.wlld.imageRecognition; import org.wlld.MatrixTools.Matrix; +import org.wlld.config.Kernel; import org.wlld.nerveEntity.SensoryNerve; import org.wlld.tools.ArithUtil; @@ -35,14 +36,6 @@ public class Operation {//进行计算 intoNerve(1, list, templeConfig.getSensoryNerves(), true, tagging); } - //卷积网络学习 - public void learning(Matrix matrix, double tag) throws Exception {//前者是图片，后者是标注 - //intoNerve(1, list, templeConfig.getSensoryNerves(), true, tagging); - intoNerveMatrix(1, matrix, - templeConfig.getConvolutionalNerveManager().getSensoryNerves() - , true, tag); - } - //图像视觉 public void look(Matrix matrix, long eventId) throws Exception { List list = convolution(matrix); @@ -73,12 +66,4 @@ public class Operation {//进行计算 sensoryNerveList.get(i).postMessage(eventId, featurList.get(i), isStudy, map); } } - - private void intoNerveMatrix(long eventId, Matrix matrix, - List sensoryNerveList - , boolean isStudy, double E) throws Exception { - for (int i = 0; i < sensoryNerveList.size(); i++) { - sensoryNerveList.get(i).postMatrixMessage(eventId, matrix, isStudy, E); - } - } } \ No newline at end of file diff --git a/src/main/java/org/wlld/imageRecognition/TempleConfig.java b/src/main/java/org/wlld/imageRecognition/TempleConfig.java index 263b1ae..67e2db2 100644 --- a/src/main/java/org/wlld/imageRecognition/TempleConfig.java +++ b/src/main/java/org/wlld/imageRecognition/TempleConfig.java @@ -1,6 +1,5 @@ package org.wlld.imageRecognition; -import org.wlld.function.ReLu; import org.wlld.function.Sigmod; import org.wlld.i.OutBack; import org.wlld.nerveCenter.NerveManager; @@ -12,7 +11,6 @@ import java.util.List; public class TempleConfig { private NerveManager nerveManager;//神经网络管理器 - private NerveManager convolutionalNerveManager;//卷积网络管理器 public double cutThreshold = 10;//切割阈值默认值 public int row = 5;//行的最小比例 public int column = 3;//列的最小比例 @@ -31,28 +29,6 @@ public class TempleConfig { } } - public NerveManager getConvolutionalNerveManager() { - return convolutionalNerveManager; - } - - public void initConvolutionalVision(boolean initPower, int weight, int height) throws Exception {//初始化卷积视觉 - //weight 是宽度 是 j 是 cloumn , height 是 高度， row 是 i - //我先要确定卷积层有几层 - int deeps = 0; - boolean isNext; - do { - height = height / 3; - weight = weight / 3; - isNext = height > 3 && weight > 3; - if (isNext) { - deeps++; - } - } while (isNext); - convolutionalNerveManager = new NerveManager(1, 1 - , 1, deeps - 1, new ReLu()); - convolutionalNerveManager.init(initPower, true); - } - public ModelParameter getModel() {//获取模型参数 return nerveManager.getModelParameter(); } diff --git a/src/main/java/org/wlld/nerveCenter/NerveManager.java b/src/main/java/org/wlld/nerveCenter/NerveManager.java index 9a9d810..e156275 100644 --- a/src/main/java/org/wlld/nerveCenter/NerveManager.java +++ b/src/main/java/org/wlld/nerveCenter/NerveManager.java @@ -168,7 +168,7 @@ public class NerveManager { List 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, hiddenNerverNub, 0, studyPoint, initPower, activeFunction, isMatrix); + OutNerve outNerve = new OutNerve(i, hiddenNerverNub, 0, studyPoint, initPower, activeFunction); //输出层神经元连接最后一层隐层神经元 outNerve.connectFathor(lastNeveList); outNevers.add(outNerve); @@ -204,8 +204,7 @@ public class NerveManager { } else { downNub = hiddenNerverNub; } - HiddenNerve hiddenNerve = new HiddenNerve(j, i + 1, upNub, downNub, studyPoint, initPower, activeFunction - , isMatrix); + HiddenNerve hiddenNerve = new HiddenNerve(j, i + 1, upNub, downNub, studyPoint, initPower, activeFunction); hiddenNerveList.add(hiddenNerve); } depthNerves.add(hiddenNerveList); diff --git a/src/main/java/org/wlld/nerveEntity/HiddenNerve.java b/src/main/java/org/wlld/nerveEntity/HiddenNerve.java index af4bd73..d53e55d 100644 --- a/src/main/java/org/wlld/nerveEntity/HiddenNerve.java +++ b/src/main/java/org/wlld/nerveEntity/HiddenNerve.java @@ -16,8 +16,8 @@ public class HiddenNerve extends Nerve { //static final Logger logger = LogManager.getLogger(HiddenNerve.class); public HiddenNerve(int id, int depth, int upNub, int downNub, double studyPoint, - boolean init, ActiveFunction activeFunction, boolean isMatrix) {//隐层神经元 - super(id, upNub, "HiddenNerve", downNub, studyPoint, init, activeFunction, isMatrix); + boolean init, ActiveFunction activeFunction) {//隐层神经元 + super(id, upNub, "HiddenNerve", downNub, studyPoint, init, activeFunction); this.depth = depth; } @@ -40,18 +40,4 @@ public class HiddenNerve extends Nerve { } // sendMessage(); } - - @Override - protected void inputMatrix(long eventId, Matrix parameter, boolean isStudy - , double E) throws Exception { - //先将输入矩阵进行卷积运算 - if (isStudy) {//初始化求和参数 - initFeatures(eventId); - } - Matrix matrix = convolution(parameter, eventId, isStudy);//这个神经元卷积结束 - if (isStudy) {//求和 - outNub = calculation(eventId); - } - sendMatrixMessage(eventId, matrix, isStudy, E);//将矩阵发到下一层神经元 - } } diff --git a/src/main/java/org/wlld/nerveEntity/Nerve.java b/src/main/java/org/wlld/nerveEntity/Nerve.java index 3260662..4c1bd4f 100644 --- a/src/main/java/org/wlld/nerveEntity/Nerve.java +++ b/src/main/java/org/wlld/nerveEntity/Nerve.java @@ -1,9 +1,7 @@ package org.wlld.nerveEntity; -import org.wlld.MatrixTools.Matrix; -import org.wlld.MatrixTools.MatrixOperation; + import org.wlld.i.ActiveFunction; -import org.wlld.test.Ma; import org.wlld.tools.ArithUtil; import java.util.*; @@ -50,21 +48,14 @@ public abstract class Nerve { } protected Nerve(int id, int upNub, String name, int downNub, - double studyPoint, boolean init, ActiveFunction activeFunction - , boolean isMatrix) {//该神经元在同层神经元中的编号 + double studyPoint, boolean init, ActiveFunction activeFunction) {//该神经元在同层神经元中的编号 this.id = id; this.upNub = upNub; this.name = name; this.downNub = downNub; this.studyPoint = studyPoint; this.activeFunction = activeFunction; - if (init) { - if (isMatrix) { - initKernel(); - } else { - initPower();//生成随机权重 - } - } + initPower(init);//生成随机权重 } public void sendMessage(long enevtId, double parameter, boolean isStudy, Map E) throws Exception { @@ -77,18 +68,6 @@ public abstract class Nerve { } } - //正向传播矩阵 - public void sendMatrixMessage(long enevtId, Matrix parameter, boolean isStudy, - double E) throws Exception { - if (son.size() > 0) { - for (Nerve nerve : son) { - nerve.inputMatrix(enevtId, parameter, isStudy, E); - } - } else { - throw new Exception("this layer is lastIndex"); - } - } - private void backSendMessage(long eventId) throws Exception {//反向传播 if (fathor.size() > 0) { for (int i = 0; i < fathor.size(); i++) { @@ -102,12 +81,6 @@ public abstract class Nerve { } - //接收矩阵参数 - protected void inputMatrix(long eventId, Matrix parameter, boolean isStudy - , double E) throws Exception {//输入 - - } - private void backGetMessage(double parameter, long eventId) throws Exception {//反向传播 backNub++; sigmaW = ArithUtil.add(sigmaW, parameter); @@ -135,6 +108,7 @@ public abstract class Nerve { double wp = ArithUtil.mul(bn, h);//编号为KEY的上层隐层神经元权重的变化值 w = ArithUtil.add(w, wp);//修正后的编号为KEY的上层隐层神经元权重 double dm = ArithUtil.mul(w, gradient);//返回给相对应的神经元 + // System.out.println("allG==" + allG + ",dm==" + dm); wg.put(key, dm);//保存上一层权重与梯度的积 dendrites.put(key, w);//保存修正结果 } @@ -157,14 +131,6 @@ public abstract class Nerve { return allReady; } - protected void initFeatures(long eventId) {//初始化九个参数和 - List list = new ArrayList<>(); - for (int i = 0; i < 9; i++) { - list.add(0.0); - } - features.put(eventId, list); - } - protected void destoryParameter(long eventId) {//销毁参数 features.remove(eventId); } @@ -183,73 +149,23 @@ public abstract class Nerve { return ArithUtil.sub(sigma, threshold); } - //进行卷积运算 - protected Matrix convolution(Matrix matrix, long eventId, boolean isStudy) throws Exception { - Matrix kernel = getKernel(); - int xn = matrix.getX(); - int yn = matrix.getY(); - int x = xn / 3;//求导后矩阵的行数 - int y = yn / 3;//求导后矩阵的列数 - Matrix myMatrix = new Matrix(x, y);//最终合成矩阵 - for (int i = 0; i < xn - 3; i += 3) {//遍历行 - for (int j = 0; j < yn - 3; j += 3) {//遍历每行的列 - //进行卷积运算 - double dm = MatrixOperation.convolution(matrix, kernel, i, j); - if (isStudy) {//如果是学习的话，拿到分块矩阵,对对应权重的值进行求和 - Matrix matrix1 = matrix.getSonOfMatrix(i, j, 3, 3); - sigma(matrix1, eventId); - } - if (dm > 0) {//存在边缘 - myMatrix.setNub(i, j, dm); - } - } - } - return myMatrix; - } - - //对每一项进行求和，只有在学习的时候执行它 - private void sigma(Matrix matrix, long eventId) throws Exception { - int n = 0; - List list = features.get(eventId); - for (int i = 0; i < 3; i++) { - for (int j = 0; j < 3; j++) { - list.set(n, ArithUtil.add(list.get(n), matrix.getNumber(i, j))); - n++; - } - } - } - - private Matrix getKernel() throws Exception { - Matrix kernel = new Matrix(3, 3); - for (int i = 1; i < 10; i++) { - double w = dendrites.get(i); - int t = i - 1; - //将权重填入卷积核当中 - int x = t / 3; - int y = t % 3; - kernel.setNub(x, y, w); - } - return kernel; - } - - private void initPower() {//初始化权重及阈值 + private void initPower(boolean init) {//初始化权重及阈值 if (upNub > 0) { Random random = new Random(); for (int i = 1; i < upNub + 1; i++) { - dendrites.put(i, random.nextDouble());//random.nextDouble() + double nub = 0; + if (init) { + nub = random.nextDouble(); + } + dendrites.put(i, nub);//random.nextDouble() } //生成随机阈值 - threshold = random.nextDouble(); - } - } - - public void initKernel() {//对卷积核进行初始化 - Random random = new Random(); - for (int i = 1; i < 10; i++) { - dendrites.put(i, random.nextDouble()); + double nub = 0; + if (init) { + nub = random.nextDouble(); + } + threshold = nub; } - //生成随机阈值 - threshold = random.nextDouble(); } public int getId() { diff --git a/src/main/java/org/wlld/nerveEntity/OutNerve.java b/src/main/java/org/wlld/nerveEntity/OutNerve.java index 2d254b7..abccab4 100644 --- a/src/main/java/org/wlld/nerveEntity/OutNerve.java +++ b/src/main/java/org/wlld/nerveEntity/OutNerve.java @@ -19,8 +19,8 @@ public class OutNerve extends Nerve { private double allE;//训练累计EK public OutNerve(int id, int upNub, int downNub, double studyPoint, boolean init, - ActiveFunction activeFunction, boolean isMatrix) { - super(id, upNub, "OutNerve", downNub, studyPoint, init, activeFunction, isMatrix); + ActiveFunction activeFunction) { + super(id, upNub, "OutNerve", downNub, studyPoint, init, activeFunction); } public void setOutBack(OutBack outBack) { @@ -54,27 +54,9 @@ public class OutNerve extends Nerve { } } - @Override - protected void inputMatrix(long eventId, Matrix parameter, boolean isStudy, double E) throws Exception { - if (isStudy) {//初始化求和参数 - initFeatures(eventId); - } - Matrix matrix = convolution(parameter, eventId, isStudy);//这个神经元卷积结束 - if (isStudy) {//求和 - outNub = calculation(eventId); - //进行反向传播 - - } else {//进入BP网络输出 - - } - } - private double outGradient() {//生成输出层神经元梯度变化 //上层神经元输入值 * 当前神经元梯度*学习率 =该上层输入的神经元权重变化 //当前梯度神经元梯度变化 *学习旅 * -1 = 当前神经元阈值变化 - //ArithUtil.sub(E, outNub) 求这个的累计平均值 - //allE = ArithUtil.add(ArithUtil.sub(E, outNub), allE); - //double avg = ArithUtil.div(allE, trainNub); return ArithUtil.mul(activeFunction.functionG(outNub), ArithUtil.sub(E, outNub)); } } diff --git a/src/main/java/org/wlld/nerveEntity/SensoryNerve.java b/src/main/java/org/wlld/nerveEntity/SensoryNerve.java index f00a1c4..8c69284 100644 --- a/src/main/java/org/wlld/nerveEntity/SensoryNerve.java +++ b/src/main/java/org/wlld/nerveEntity/SensoryNerve.java @@ -13,7 +13,7 @@ import java.util.Map; public class SensoryNerve extends Nerve { public SensoryNerve(int id, int upNub) { - super(id, upNub, "SensoryNerve", 0, 0.1, false, null, false); + super(id, upNub, "SensoryNerve", 0, 0.1, false, null); } public void postMessage(long eventId, double parameter, boolean isStudy, Map E) throws Exception {//感知神经元输出 @@ -21,10 +21,6 @@ public class SensoryNerve extends Nerve { sendMessage(eventId, parameter, isStudy, E); } - public void postMatrixMessage(long enevtId, Matrix parameter, boolean isStudy, double E) throws Exception { - sendMatrixMessage(enevtId, parameter, isStudy, E); - } - @Override public void connect(List nerveList) {//连接第一层隐层神经元 super.connect(nerveList); diff --git a/src/main/java/org/wlld/test/Test.java b/src/main/java/org/wlld/test/Test.java index d38fde3..0a39315 100644 --- a/src/main/java/org/wlld/test/Test.java +++ b/src/main/java/org/wlld/test/Test.java @@ -7,7 +7,10 @@ package org.wlld.test; */ public class Test { public static void main(String[] args) { - int a = 1 % 3; - System.out.println(a); + + } + + public static void test1() { + } }