Neural network
WordDictionary
-Dictionary:Trie
-rankWordsBySfDiff()+outputTrainData(int):TrainData+update(List<String>)
Trie
-root:TrieNode
+Trie():void+insert(String):void+has(String):boolean+preTraverse(TrueNode) void+countPrefix(String):int
contains1...*
generate
Feature extraction
Neuron
- m_ForwardInputValue：float- m_ForwardOutputValue: float- m_BackWardInputValue: floatm_ackwardOutputValue: float
+ Neuron() :void-forwardSigmoid(float): float-logS(float): float-logSDerivative(float): float-tanhS(float ): float-tanhSDerivative(float): float-backwardPropagate(float) :float+getForwardInputValue(): float+setForwardInputValue():void+getForwardOutputValue():float+setForwardOutputValue(float): void+getBackwardInputValue() : float+setBackwardInputValue(float) :void+getBackwardOutputValue():float+setBackwardOutputValue(float): void
input
Word
+ name:String+ gameFile_num:int+ ungameFile_num:int-sf_diff:double
+Word(String):void+addGameFIleNum():void+addUnGameFIleNum():void+calcSfDiffVal():double
DataNode
-InputData:List<Float>-outputData:int
Layer
File
+wordTree:Trie<Node>-fileName:string-fileType:Enum
+ addWord(String):void- findWord(String):void-getFileWords:List<String)
contains2
Net
-mInputCount:int-mHiddenCount:int-mOutputCount:int-mInputNodes:List<Neuron>-mHiddenNodes: List<Neuron> -mOutputNodes: List<Neuron> -mInputHiddenWeight:float[][]-mHiddenOutputWeight:float[][] -trainNodes:List<DataNode>
TrieNode
-num:int-son: TrieNode[] -isEnd: boolean-word:Word;
+TrieNode(word):void+getWord():Word
