人工智能主觀貝葉斯分析實(shí)驗(yàn)

1、人工智能實(shí)驗(yàn)報(bào)告西安交大-、實(shí)驗(yàn)?zāi)康?1) 學(xué)習(xí)了解java編程語(yǔ)言，掌握基本的算法實(shí)現(xiàn)；(2) 深入理解貝葉斯理論和不確定性推理理論；(3) 學(xué)習(xí)運(yùn)用主觀貝葉斯公式進(jìn)行不確定推理的原理和過程二、實(shí)驗(yàn)題目用java語(yǔ)言實(shí)現(xiàn)運(yùn)用主觀貝葉斯公式進(jìn)行不確定性推理的過程：根據(jù)初始證 據(jù)E的概率P(E)及LS LN的值，把H的先驗(yàn)概率P(H)更新為后驗(yàn)概率P(H/E) 或者P(H/E)。要求如下：(1) 充分考慮各種證據(jù)情況：證據(jù)肯定存在、證據(jù)肯定不存在、觀察與證據(jù) 無(wú)關(guān)、其他情況；(2) 考慮EH公式和CP公式兩種計(jì)算后驗(yàn)概率的方法；(3) 給出EH公式的分段線性插值圖；三、實(shí)驗(yàn)原理1 、知識(shí)的不確定

2、性在主觀貝葉斯方法中，只是是如下形式的產(chǎn)生式規(guī)則表示:IF ETHEN(LS,LN) H (P(H)LS是充分性度量。其定義為：LS=P(E|H)/P(E| ?H)。LN是必要性度量，其定義為：LN=P(?E|H)/P( ?E| ?H)=(1-P(E|H)/(1-P(E|?H)。2、證據(jù)不確定時(shí)的計(jì)算公式當(dāng) 0< P(E/S) ： P(E)當(dāng)P(E)乞 P(E/S)乞 1l P(H)-P(H/E)_l P(H/ E)*P(E/S)P(H/S)二P(E)P(H) P(H/E)_P(H)s) _P(E)1-P(E)四、實(shí)驗(yàn)代碼public class bayes extends JFrame

3、 implements ActionListener JPanel panel= newJPanel();JLabel ph= new JLabel("P(H)");JTextField PH= new JTextField("",3);JLabel pe= new JLabel("P(E)");JTextField PE= new JTextField("",3);JLabel ls= new JLabel("LS");JTextField LS= new JTextField("&

4、quot;,3);JLabel ln= new JLabel("LN");JTextField LN= new JTextField("",3);Button compute= newButton("COMPUTE");staticdoublet_ph;staticdoublet_pe;staticdoublet_ln ;staticdoublet_ls ;staticdoubleph_e;/P(E/S)=0時(shí) PHSstaticdoublephe;/P(E/S)=1時(shí) PHSpublicsebayes()tLayout(new Bo

5、rderLayout();panel.setLayout( newFlowLayout();panel.add(ph);panel.add(PH);panel.add(pe);panel.add(PE);panel.add(ln);panel.add(LN);panel.add(ls);panel.add(LS);this .add(panel);compute.addActionListener( this );this .add(compute,BorderLayout.SOUT)H;public static void main(String args)bayes a= new baye

6、s();a.setSize(400,250);a.setVisible( true );a.setDefaultCloseOperatio n( EXIT_ON_CLOSEOverridepublic void acti on Performed(Acti on Eve nt argO) / TOD8uto-ge nerated method stubt_ph =new Double(PH.getText();t_pe =new Double(PE.getText();t_ls =n ew Double(LS.getText();t_ln =new Double(LN.getText();ph

7、_e=t_ln *t_ph /( t_ln -1)* t_ph+1);phe=t_ls *t_ph/( t_ls -1)* t_ph +1);display c= new display();class draw extends JPanelpublic void paint(Graphics g)super .paint(g);g.drawLine(50, 350, 350, 350);g.drawL in e(50, 50, 50, 350 );g.drawLine(50, 350-( int )(bayes. ph_e*300),50+( int )(bayes. t_pe *300),

8、350-( int )(bayes. t_ph *300);g.drawLine(50+(int )(bayes. t_pe *300),350-( int )(bayes. t_ph *300),350,350-( int )(bayes. phe*300);class displayextends JFramepublic display()draw b= new draw();this .add(b);this .setDefaultCloseOperation(JFrame.EXIT_ON_CLOSEthis .setVisible( true );this .setSize(400,400);五、實(shí)驗(yàn)結(jié)果輸入初始值：圖像結(jié)果顯示:六、實(shí)驗(yàn)總結(jié)由