类 NerveManager
java.lang.Object
org.wlld.nerveCenter.NerveManager
-
构造器概要
构造器构造器说明NerveManager(int sensoryNerveNub, int hiddenNerveNub, int outNerveNub, int hiddenDepth, ActiveFunction activeFunction, double studyPoint, int rzType, double lParam, int coreNumber) 初始化神经元参数 -
方法概要
修饰符和类型方法说明voidinit(boolean initPower, boolean isShowLog, boolean isSoftMax) 初始化voidinitImageNet(int step, int kernLen, int xSize, int ySize, boolean isSoftMax, boolean isShowLog, double convStudyPoint, ActiveFunction convFunction) 初始化卷积层神经网络voidinsertConvModel(ModelParameter modelParameter) voidinsertDnnModel(ModelParameter modelParameter)
-
构造器详细资料
-
NerveManager
public NerveManager(int sensoryNerveNub, int hiddenNerveNub, int outNerveNub, int hiddenDepth, ActiveFunction activeFunction, double studyPoint, int rzType, double lParam, int coreNumber) throws Exception 初始化神经元参数- 参数:
sensoryNerveNub- 输入神经元个数hiddenNerveNub- 隐层神经元个数outNerveNub- 输出神经元个数hiddenDepth- 隐层深度activeFunction- 激活函数studyPoint- 线性分类器学习率rzType- 正则函数lParam- 正则系数coreNumber- 并行计算核心数- 抛出:
Exception- 如果参数错误则抛异常
-
-
方法详细资料
-
getConvModel
- 抛出:
Exception
-
getDnnModel
- 抛出:
Exception
-
insertConvModel
- 抛出:
Exception
-
insertDnnModel
-
getSensoryNerves
-
initImageNet
public void initImageNet(int step, int kernLen, int xSize, int ySize, boolean isSoftMax, boolean isShowLog, double convStudyPoint, ActiveFunction convFunction) throws Exception 初始化卷积层神经网络- 参数:
step- 卷积步长 建议为2kernLen- 卷积核大小 建议为3xSize- 检测窗口行高ySize- 检测窗口行宽isSoftMax- 最后一层是否用softMax激活isShowLog- 是否打印学习参数convStudyPoint- 卷积层学习率convFunction- 卷积层激活函数- 抛出:
Exception
-
init
-